the influence of knowledge sharing on motivation ... - Semantic Scholar

0 downloads 0 Views 2MB Size Report
Organisations wishing to enhance their return on investment from training must ...... lower than the initial 82 percent return rate, it is still a high rate of return for ...... Di bawah ini adalah maklumat penting yang berkaitan dengan soal selidik yang dilampirkan. ... Arahan : Bagi Nombor 1, sila berikan nama latihan yang dihadiri.
THE INFLUENCE OF KNOWLEDGE SHARING ON MOTIVATION TO TRANSFER TRAINING: A MALAYSIAN PUBLIC SECTOR CONTEXT

A thesis submitted in fulfilment of the requirement for the degree of DOCTOR OF PHILOSOPHY by

Shahril Bin Baharim

School of Management Faculty of Business and Law Victoria University Melbourne, Australia February 2008

ABSTRACT Organisations wishing to enhance their return on investment from training must understand the variables associated with transfer of training so that they can promote those which enable transfer and intervene to limit those which inhibit transfer. In the international literature on training transfer, researchers and practitioners have acknowledged that transfer of training will occur only when trainees have the desire or motivation to transfer training to the job. In Malaysia, despite increasing investment in public sector training, there has been very little research on transfer of training. This thesis contributes to a greater understanding of transfer of training variables and how they affect trainees’ motivation to transfer their training. Further, as the role of training has progressively changed from a focus on programs to a broader focus on learning, creating and sharing knowledge, this thesis tests the hypothesis that knowledge sharing behaviour influences a trainee’s motivation to transfer their training. Using a research framework constructed from an adaptation of two key Human Resource Development models (Holton 1996; Holton et al. 2000) and the theory of planned behaviour (Ajzen 1991), this thesis explores the contention that trainees’ motivation to transfer training is influenced by a number of secondary influence variables, expected utility variables, transfer climate variables, enabling variables and ability variables as well as the variables associated with sharing behaviour. Through a questionnaire administered to 437 government employees attending training programs in the National Institute of Public Administration, a central training organisation for government employees in Malaysia, the thesis created an empirical database from which to study the phenomenon of transfer of training. This work culminated in the development of a structural model for motivation to transfer training which incorporates knowledge sharing behaviour and extends our understanding of the operation of the precursors to motivation to transfer. The findings of this thesis impact on HRD functions in the Malaysian public sector at two broad levels: pre training and post training. The thesis makes a contribution to both HRD practice by detailing the sorts of HRD activities which will enhance transfer of training and secondly, makes a contribution to theory through the creation of a new model of motivation to transfer training which features knowledge sharing behaviour.

ii

TABLE OF CONTENTS ABSTRACT ii DECLARATION xiv PUBLICATIONS xv ACKNOWLEDGEMENTS xvi

CHAPTER 1: INTRODUCTION 1.1 Background to the Research

1

1.2 Research on Training and Its Transfer in Malaysia

3

1.3 Purpose and Research Questions

5

1.4 Justification for the Research

12

1.4.1 Significance of the Thesis

12

1.5 Research Approach

13

1.6 Outline of the Thesis

13

1.7 Research Limitations and Assumptions

15

1.8 Summary

16

CHAPTER 2: TRANSFER OF TRAINING: A REVIEW OF THE RESEARCH LITERATURE 2.1 Introduction

18

2.2 Training and Transfer of Training

19

2.3 Motivation to Transfer Training

22

2.3.1 The Kirkpatrick Model

22

2.3.2 The Learning Transfer System Inventory (LTSI)

26 iii

2.4 The Human Resource Development (HRD) Evaluation Research and Measurement Model

31

2.4.1 Influences on Motivation to Transfer

33

2.4.2 Influences on Motivation to Learn

39

2.4.3 Influences on Learning Outcomes

42

2.4.4 Influences on Individual Performance

44

2.5 Knowledge Sharing and Its Benefits

46

2.6 The Theory of Planned Behaviour

49

2.7 Summary

51

CHAPTER 3: THE RESEARCH FRAMEWORK AND METHODOLOGY 3.1 Introduction

53

3.2 The Conceptual Framework

53

3.2.1 The Research Questions 3.3 Methodology

58 62

3.3.1 Questionnaire Design

62

3.3.2 Sample and Data Collection

71

3.3.3 Analysis Strategy

74

3.4 Summary

90

CHAPTER 4: CONSTRUCT VALIDITY AND RELIABILITY 4.1 Introduction

92

4.2 Construct Assessment

92

4.2.1 The Learner Readiness Construct

95

4.2.2 The Performance-Self Efficacy Construct

97 iv

4.2.3 The Motivation to Transfer Construct

99

4.2.4 The Transfer Effort-Performance Expectations Construct

101

4.2.5 The Performance-Outcomes Expectations Construct

104

4.2.6 The Feedback Construct

106

4.2.7 The Peer Support Construct

109

4.2.8 The Supervisor Support Construct

111

4.2.9 The Openness to Change Construct

113

4.2.10 The Personal Outcomes-Positive Construct

115

4.2.11 The Personal Outcomes-Negative Construct

118

4.2.12 The Supervisor Sanctions Construct

120

4.2.13 The Personal Capacity for Transfer Construct

123

4.2.14 The Opportunity to Use Construct

125

4.2.15 The Content Validity Construct

127

4.2.16 The Transfer Design Construct

130

4.2.17 The Sharing Behaviour Construct

132

4.2.18 The Intention to Share Construct

134

4.2.19 The Attitude Toward Knowledge Sharing Construct

137

4.2.20 The Subjective Norm Toward Knowledge Sharing Construct

140

4.2.21 The Perceived Behavioural Control Toward Knowledge Sharing Construct

142

4.3 Discussion

145

4.4 Summary

148

v

CHAPTER 5: HYPOTHESIS TESTING: RESULTS AND DISCUSSION 5.1 Introduction

150

5.2 Distribution of Respondents

150

5.2.1 Distribution of Respondents by Training Types

151

5.2.2 Distribution of Respondents by Gender

152

5.2.3 Distribution of Respondents by Age

152

5.2.4 Distribution of Respondents by Level of Education

153

5.2.5 Distribution of Respondents by Work Experience

153

5.2.6 Distribution of Respondents by Position of Employment

154

5.3 Hypothesis Testing

155

5.3.1 Hypothesis Testing for Research Question One

155

5.3.2 The Relationship Between the Transfer of Training Variables and the Type of Training Undertaken

157

5.3.3 Hypothesis Testing for Research Question Two

165

5.3.4 The Relationship Between the Transfer of Training Variables and Trainee Demographics

173

5.4 Discussion

176

5.5 Summary

177

CHAPTER 6: HYPOTHESIS TESTING: RESULTS AND DISCUSSION (PART 2) 6.1 Introduction

178

6.2 Hypothesis Testing

178

6.2.1 Hypothesis Testing for Research Question Three

179

6.2.2 The Key Significant Predictors of One’s Motivation to Transfer Training

187

6.2.3 Hypothesis Testing for Research Question Four

191 vi

6.2.4 The Relationships Between Intention to Share, Sharing Behaviour and Motivation to Transfer

192

6.2.5 Hypothesis Testing for Research Question Five

193

6.2.6 The Significant Predictors of Intention to Share

195

6.2.7 Hypothesis Testing for Research Question Six

196

6.2.8 The Direct and Indirect Relationships Between Sharing Behaviour and Motivation to Transfer

209

6.3 The Evolution of the Final Structural Model

213

6.4 Discussion

215

6.5 Summary

216

CHAPTER 7: IMPLICATIONS FOR HRD PRACTISE, THEORY, RESEARCH LIMITATIONS, GENERALISABILITY AND DIRECTIONS FOR FUTURE RESEARCH 7.1 Introduction

217

7.2 Implications for HRD Practice

217

7.2.1 The Importance of Diagnosing Transfer of Training

218

Variables 7.2.2 The Influence of Sharing Behaviour on

225

Motivation to Transfer 7.3 Implications for HRD Theory

233

7.4 Generalisability of the Study

236

7.5 Research Limitations

237

7.6 Future Research

238

7.7 Summary

240

vii

CHAPTER 8: SUMMARY AND CONCLUSIONS 8.1 Introduction

241

8.2 Summary of the Findings for HRD Practice

242

8.3 Summary of the Findings for HRD Theory

244

REFERENCES

246-262

APPENDICES

263-348

Appendix A: Main Study Questionnaire (English and Bahasa MalaysiaVersions) Appendix B: List of Items Generated for Each Construct Appendix C: Letter of Approval for the Pilot Study and Main Study Data Collection Appendix D: Letter of Approval from Ethics Committee, Victoria University Appendix E: The Number of Distributed and Returned Questionnaire Appendix F: The Summary of Main Study Questionnaire Appendix H: Normal Probability Plot for the Assumption of Normality Appendix I: Scatter Plot for the Assumption of Linearity, Homoscedasticity and Independence of Residual Appendix J: Item-to-Total Correlations, Inter-Item Correlations and Cronbach Alpha for Each Construct Appendix K: Construct Reliability and Variance Extracted Workings Appendix L1: Multivariate Analysis of Variance Across Training Types Appendix L2: Multivariate Analysis of Variance Across Gender Appendix L3: Multivariate Analysis of Variance Across Age Appendix L4: Multivariate Analysis of Variance Across Level of Education Appendix L5: Multivariate Analysis of Variance Across Work Experience Appendix L6: Multivariate Analysis of Variance Across Position of Employment Appendix N: Regression Coefficient and Measurement Error Variance Workings

viii

LIST OF TABLES Table 1.1: The Definitions of the Variables Table 2.1: The 16 Factors of the LTSI Which Affect Transfer of Training Table 3.1: Factors removed from the original HRD models Table 3.2: The Statement of Hypotheses Table 3.3: Experts Comments Table 3.4: Results of the Item Analysis Table 3.5: Correlation Matrix Table 4.1: Learner Readiness Principle Component Analysis Table 4.2: Fit Indices for Learner Readiness Construct Table 4.3: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Learner Readiness Construct Table 4.4: Performance-Self Efficacy Principle Component Analysis Table 4.5: Fit Indices for Performance-Self Efficacy Conctruct Table 4.6: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Performance-Self Efficacy Construct Table 4.7: Motivation to Transfer Principle Component Analysis Table 4.8: Fit Indices for Motivation to Transfer Construct Table 4.9: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Motivation to Transfer Construct Table 4.10: Transfer Effort-Performance Expectations Principle Component Analysis Table 4.11: Fit Indices for Transfer Effort-Performance Expectations Construct Table 4.12: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Transfer Effort-Performance Expectations Construct Table 4.13: Performance-Outcomes Expectations Principle Component Analysis Table 4.14: Fit Indices for Performance-Outcomes Expectations Construct Table 4.15: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Performance-Outcomes Expectations Construct Table 4.16: Feedback Principle Component Analysis Table 4.17: Fit Indices for Feedback Construct Table 4.18: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Feedback Construct Table 4.19: Peer Support Principle Component Analysis Table 4.20: Fit Indices for Peer Support Construct Table 4.21: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Peer Support Construct Table 4.22: Supervisor Support Principle Component Analysis Table 4.23: Fit Indices for Supervisor Support Construct Table 4.24: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Supervisor Support Construct Table 4.25: Openness to Change Principle Component Analysis Table 4.26: Fit Indices for Openness to Change Construct

ix

Table 4.27: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Openness to Change Construct Table 4.28: Personal Outcomes-Positive Principle Component Analysis Table 4.29: Fit Indices for Personal Outcomes-Positive Construct Table 4.30: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Personal Outcomes-Positive Construct Table 4.31: Personal Outcomes-Negative Principle Component Analysis Table 4.32: Fit Indices for Personal Outcomes-Negative Construct Table 4.33: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Personal Outcomes-Negative Construct Table 4.34: Supervisor Sanctions Principle Component Analysis Table 4.35: Fit Indices for Supervisor Sanctions Construct Table 4.36: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Supervisor Sanctions Construct Table 4.37: Personal Capacity for Transfer Principle Component Analysis Table 4.38: Fit Indices for Personal Capacity for Transfer Construct Table 4.39: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Personal Capacity for Transfer Construct Table 4.40: Opportunity to Use Principle Component Analysis Table 4.41: Fit Indices for Opportunity to Use Construct Table 4.42: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Opportunity to Use Construct Table 4.43: Content Validity Principle Component Analysis Table 4.44: Fit Indices for Content Validity Construct Table 4.45: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Content Validity Construct Table 4.46: Transfer Design Principle Component Analysis Table 4.47: Fit Indices for Transfer Design Construct Table 4.48: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Transfer Design Construct Table 4.49: Sharing Behaviour Principle Component Analysis Table 4.50: Fit Indices for Sharing Behaviour Construct Table 4.51: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Sharing Behaviour Construct Table 4.52: Intention to Share Principle Component Analysis Table 4.53: Fit Indices for Intention to Share Construct Table 4.54: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Intention to Share Construct Table 4.55: Attitude Toward Knowledge Sharing Principle Component Analysis Table 4.56: Fit Indices for Attitude Toward Knowledge Sharing Construct Table 4.57: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Attitude Toward Knowledge Sharing Construct Table 4.58: Subjective Norm Toward Knowledge Sharing Principle Component Analysis Table 4.59: Fit Indices for Subjective Norm Toward Knowledge Sharing Construct Table 4.60: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Subjective Norm Toward Knowledge Sharing Construct Table 4.61: Perceived Behavioural Control Principle Component Analysis x

Table 4.62: Fit Indices for Perceived Behavioural Control Construct Table 4.63: Descriptive Statistics, Cronbach Alpha, Construct Reliability and Variance Extracted for Perceived Behavioural Control Construct Table 5.1: Distribution of Respondents by Training Types Table 5.2: Distribution of Respondents by Gender Table 5.3: Distribution of Respondents by Age Table 5.4: Distribution of Respondents by Level of Education Table 5.5: Distribution of Respondents by Work Experience Table 5.6: Distribution of Respondents by Position of Employment Table 5.7: The Variables That Differed Across Gender, Age, Level of Education, Work Experience and Position of Employment Table 6.1: Model Summary Table 6.2: ANOVA Table 6.3: Coefficients-Secondary Influence Variables Table 6.4: Model Summary Table 6.5: ANOVA Table 6.6: Coefficients-Expected Utility Variables Table 6.7: Model Summary Table 6.8: ANOVA Table 6.9: Coefficients-Transfer Climate Variables Table 6.10: Model Summary Table 6.11: ANOVA Table 6.12: Coefficients-Enabling Variables Table 6.13: Model Summary Table 6.14: ANOVA Table 6.15: Coefficients-Ability Variables Table 6.16: Summary of Results for Research Question Three Table 6.17: Model Summary Table 6.18: ANOVA Table 6.19: Coefficients-Antecedents of Intention to Share Table 6.20: Fit Indices for the Hypothesised Structural Model Table 6.21: Results of the Hypothesised Structural Model Table 6.22: Fit Indices for the Re-Specified Model 1 Table 6.23: Results of the Relationships for Re-Specified Model 1 Table 6.24: Fit Indices for the Final Structural Model Table 6.25: Results of the Relationships for the Final Structural Model Table 7.1: Pre-training and Post-Training Variables for HRD Functions

xi

LIST OF FIGURES Figure 1.1: The Conceptual Framework Figure 2.1: The Kirkpatrick Four Level Evaluation Model Figure 2.2: The LTSI Conceptual Evaluation Model Figure 2.3: The Learning Transfer System Inventory (LTSI) Figure 2.4: The HRD Evaluation Research and Measurement Model Figure 2.5: Evolution’s of Training Role Figure 2.6: The Theory of Planned Behaviour Figure 3.1: The Conceptual Framework Figure 3.2: The Framework for Developing Questionnaire Figure 3.3: Simplified Structural Model Figure 4.1: Measurement Model-Learner Readiness Construct Figure 4.2: Measurement Model-Performance-Self Efficacy Construct Figure 4.3: Measurement Model-Motivation to Transfer Construct Figure 4.4: Measurement Model-Transfer Effort- Performance Expectations Construct Figure 4.5: Measurement Model-Performance-Outcomes Expectations Construct Figure 4.6: Measurement Model-Feedback Construct Figure 4.7: Measurement Model-Peer Support Construct Figure 4.8: Measurement Model-Supervisor Support Construct Figure 4.9: Measurement Model-Openness to Change Construct Figure 4.10: Measurement Model-Personal Outcomes-Positive Construct Figure 4.11: Measurement Model-Personal Outcomes-Negative Construct Figure 4.12: Measurement Model-Supervisor Sanctions Construct Figure 4.13: Measurement Model-Personal Capacity for Transfer Construct Figure 4.14: Measurement Model-Opportunity to Use Construct Figure 4.15: Measurement Model-Content Validity Construct Figure 4.16: Measurement Model-Transfer Design Construct Figure 4.17: Measurement Model-Sharing Behaviour Construct Figure 4.18: Measurement Model-Intention to Share Construct Figure 4.19: Measurement Model-Attitude Toward Knowledge Sharing Construct Figure 4.20: Measurement Model-Subjective Norm Toward Knowledge Sharing Construct Figure 4.21: Measurement Model-Perceived Behavioural Control Toward Knowledge Sharing Construct Figure 6.1: Hypothesised Structural Model Figure 6.2: The Results for the Hypothesised Structural Model Figure 6.3: The Results for the Re-Specified Model 1 Figure 6.4: The Results for the Final Structural Model

xii

xiii

DECLARATION

“I, SHAHRIL BIN BAHARIM, declare that the PhD thesis entitled ‘The Influence of Knowledge Sharing on Motivation to Transfer Training. A Malaysian Public Sector Context’, is no more than 100,000 words in length, exclusive of tables, figures, appendices and references. This thesis contains no material that has been submitted previously, in whole or in part, for the award of any other academic degree or diploma. Except where otherwise indicated, this thesis is my own work”.

Signature: ________________________ Date: _______________

xiv

PUBLICATIONS

The following are a list of publications derived from this research.

Baharim, S & Van Gramberg, B 2005, ‘The influence of knowledge sharing on transfer of training: a proposed research strategy’, in: Proceedings of the Association of Industrial Relations Academic of Australia and New Zealand Conference, 9-11 February 2005, Sydney, Australia. Baharim, S & Van Gramberg, B 2006, ‘The influence of knowledge sharing on transfer of training: a proposed research strategy’, The ICFAI Journal of Knowledge Management, vol.4, no.1, pp. 47-58. Baharim, S & Van Gramberg, B 2007, ‘The relationship between knowledge sharing and transfer of training: a conceptual model’, in Lahiri, K (eds.), ICFAI University Press. Baharim, S & Van Gramberg, B 2007, ‘Factors affecting motivation to transfer. A Malaysian Public Sector Study’, School of Management, Victoria University Working Paper.

xv

ACKNOWLEDGEMENTS At last…I have made it! What an amazing learning journey it has been. First, I would like to thank to my wonderful family, especially to my loving wife Surina who had sacrificed her time and career accompanying me in Melbourne to undertake my PhD. Her encouragement and never ending flow of moral support had given me back the confidence I so badly needed but had lost. Also to all my children Aliya, Adam, Sarah and Alif….they are all under 10 years of age, they haven’t got any idea what this task is all about but never doubted their dad will succeed. My profound thank to my parents, brothers and sisters for their never ending encouragement. Second, I would like to express my sincere appreciation to my supervisor Associate Professor Dr Bernadine Van Gramberg who took me on as her student. Her guidance, assistance, patience and support over the long years of research work are very invaluable especially during the frustrating times. I found that such supervision has significantly contributed to this research. My appreciation also goes to Richard Gough for his assistance and suggestions in statistics. I would also like to thank to Associate Professor Patrick Foley for his generous time with AMOS and structural equation modeling. I will never forget the hospitality he and his wife have shown me. Third, I would be remiss if I did not address the following individuals who helped me during the pilot and main study data collection. They are: Saadiah Yaakob (Accountant General’s Department), Amiruddin Muhammed (Ministry of Finance) and Abdul Latif Ahmad (Accountant General’s Department). Forth, I am deeply indebted to Y. Bhg. Datuk Siti Maslamah Osman (Accountant General 2000-2003), who gave me her support for a study leave. I owe her a special debt of gratitude. A special thanks to the present Accountant General Dato’ Mohd Salleh Mahmud who inspired and supported me to come to study in Australia. My special thanks also goes to Tn Hj. Ab. Gani Haron (Deputy Accountant General-Operation), Teh Ben Chu (Director EGAG) and Tn Hj. Mohsin Hussein (Director BPAS). My sincere apologies to others whose names I have not mentioned. Finally, my profound thanks to all my PhD Colleagues, especially to Tafir and his wife for providing my wife and my children with temporary stay during the last phase of my PhD. We will never forget the kindness he and her wife had shown to us.

xvi

For my wife Surina and my children Aliya, Adam, Sarah and Alif

xvii

CHAPTER 1 INTRODUCTION

1.1 Background to the Research The government of Malaysia has a vision for the country to become a fully developed nation by the year 2020 (Mohamed 2003). In order to achieve this, Malaysia requires a knowledgeable and skilful workforce to compete successfully in meeting the challenges ahead. In this regard, developing human capital is a top priority of the Malaysian government (Hashim 2001). This is evidenced by a recent government announcement of an allocation of RM 33.4 billion in the 2007 national budget to further strengthen the education and training system (Ahmad Badawi 2006). The public sector plays a vital role as an enabler and facilitator of private sector initiatives by providing efficient delivery systems and a customer-focused service. These roles lie with the approximately 1.3 million government employees working in the public sector in positions from clerical to top executives (Public Service Department of Malaysia 2006). In order to ensure that these employees are equipped with the necessary knowledge and skills, the Malaysian government has given particular attention to workplace training as a tool for improving employees’ job performance. This has included programs such as management training (for example, human resource management; strategic management; financial management), computer training (for example, visual basic; database management) and general training (for example, writing skills; better spoken English) (National Institute of Public Administration Malaysia 2005). Key to the success of training program initiatives is the extent to which trainees use their training on the job. Over the years, researchers and practitioners have acknowledged that transfer of training must occur before learning can lead to an improvement in an individual’s job performance (Holton 1996; Holton, Bates & 1

Ruona 2000). This, in turn, represents the essence of return on investment (ROI) of training. Transfer of training has been defined as the degree to which trainees apply the knowledge and skills gained in training to their job (Ford & Weissbein 1997; Tannenbaum & Yulk 1992; Wexley & Latham 1991). The acquisition of knowledge and skills gained in training is of little value if the new characteristics are not taken back to the job setting or are not maintained over time (Bates 2003; Kozlowski & Salas 1997). Surprisingly, it has been reported that a mere 10 percent of the investment in training is returned in performance improvement (Garavaglia 1993). Despite the reported problems in training transfer in the research literature, workplace training is still viewed as a primary strategy by organisations to gain a competitive advantage. This is because the goal of training is for employees to master the knowledge and skills learned, and this in turn, is argued as being critical for successful job performance (Goldstein 1992; Noe 2005; Wexley & Latham 1991). It has been noted that transfer of training will occur when trainees have the desire or motivation to use the knowledge and skills learned in their training (Baldwin & Ford 1988; Noe 1986; Noe & Schmitt 1986; Wexley & Latham 1991). However, little is known about factors that could affect trainees’ motivation to transfer training to the job (Seyler, Holton, Bates, Burnett & Carvalho 1998; Tannenbaum & Yulk 1992). Clearly, a better understanding the factors that enhance trainees’ use of their learned skills and knowledge on the job would be valuable in determining how to motivate trainees to use the knowledge and skills so that the organisation is benefited. In the light of the importance of investment in training as a strategy for the Malaysian government (and indeed for many organisations), this thesis aims to identify the key contributing factors to transfer of training. This Chapter presents an overview of the thesis, commencing with a discussion of the contention that transfer of training and in particular, one’s motivation to transfer one’s learning, are key components to productive training program design and implementation and represent factors which will contribute to ROI of training costs. The Chapter then moves to describe the

2

research questions driving the thesis and outlines briefly, the methodology and limitations of the research.

1.2 Research on Training and its Transfer in Malaysia Despite a small body of literature on employee training in Malaysia, there has been very little research on transfer of training. Of the variety of studies covering different aspects of training, one observes a general agreement that a lack of understanding of training needs assessment and training evaluation inhibits human resource development (HRD) initiatives in that country. The widespread absence of these practices may have contributed to a belief that ROI on training investment is itself, illusive. For example, Zakaria and Rozhan (1993) examined the HRD practices in the manufacturing sector in Malaysia and found that despite 44 percent of surveyed firms conducting formal training, a lack of expertise amongst training managers meant that 23 percent of them did not first conduct a training needs assessment. In another study of 54 Malaysian manufacturing firms and 46 service sector firms, Poon and Othman (2000) found that although the organisations developed basic processes such as training needs assessment and training evaluation, implementation of these processes was poorly handled. Further, training needs assessment was found to be based on past data such as job content or company records to identify training needs (rather than current audits of skills gaps, for instance). Likewise, training evaluation relied on rating sheets handed out at the end of training programs to trainees and were thus highly subjective and one dimensional. A study by Saiyadain (1995) examining attitudes towards training found that sponsoring managers (i.e., those managers directly concerned with the training activities) from smaller sized companies had relatively more negative perceptions to training than those in larger organisations. Further, training in those smaller organisations was given low priority because of the managers’ inability to see tangible benefits of training. This was confirmed by Hashim (2001) who found that

3

training evaluation was overwhelmingly reliant on reactive measures such as trainee feedback and observation rather than on other ROI measures. A related study by Saiyadain and Ali (1995) found that measurement of training effectiveness was inconsistent in Malaysian firms and that most Malaysian managers did not have formal education in management, a fact which itself, impacted negatively on their training and development efforts. Only in one study (Hameed & Analoui 1999), was the problem of transfer in management training and development examined and this was situated in the manufacturing sector. The study revealed that among the inhibiting transfer factors in the workplace (to transfer of training occurring) were the different perceptions and understanding between staff and management; the presence of a conflicting environment (training and workplace); a lack of recognition; and constraints relating to company’s policies and procedures. Although the studies described above have contributed greatly to the collective knowledge in managerial training in a Malaysian setting, they represent a very small body of research which has had little impact on HRD practice. This was confirmed by the survey of the literature on training and development of managers in Malaysia by Saiyadain and Ali (1995) who indicated that there was a dearth of published empirical material on managerial training. Since then, the study by Hashim (2001) marks the only recent contribution to the field. HRD researchers have acknowledged that the role of training has changed from a program focus to a broader focus on learning and creating and sharing knowledge (Martocchio & Baldwin 1997; Noe 2005; Noe, Hollenbeck, Gerhart & Wright 2004). Employees are expected to acquire new skills and knowledge, apply them on the job and share this information with fellow workers (Noe 2005). However, in the context of the Malaysian public sector, research investigating how knowledge sharing could play its role in transfer of training has been neglected. This thesis, therefore, sets itself squarely in the research gap with the aim of contributing to the HRD discipline

4

in Malaysia through a comprehensive study of factors which promote transfer of training in the Malaysian public sector.

1.3 Purpose and Research Questions Research suggests that transfer factors vary across organisations, training types and trainees’ demographics and they vary between being strong and weak with respect to their effect on transfer itself (Chen 2003; Donovan, Hannigan & Crowe 2001; Holton, Chen & Naquin 2003; Yamnill 2001). This means that organisations wishing to enhance ROI from training must understand all the transfer factors and intervene to remove factors that inhibit transfer. Key to this discussion is that researchers and practitioners have acknowledged that transfer of training will occur only when trainees have the desire or motivation to transfer their training on the job (Baldwin & Ford 1988; Noe 1986; Noe & Schmitt 1986; Wexley & Latham 1991). It follows that organisations would be well advised to identify the factors that enhance trainees’ motivation to transfer their training particularly as the goal of most training is for trainees to master the knowledge and skills required to perform well in their daily job (Noe 2005). This thesis contributes to a greater understanding of the nature of the motivational factors which underpin trainees’ desire to transfer what they have learned into their everyday jobs. The thesis also tests the hypothesis that knowledge sharing behaviour also influences a trainee’s motivation to transfer their training. Preliminary evidence on the concept of knowledge sharing behaviour (Noe et al. 2004) demonstrates that it plays some role in transfer of training by acting through trainee motivational factors. Two key models were used to explore the areas of motivation and knowledge sharing. First, using the Human Resource Development Evaluation Research and Measurement Model (the HRD model) developed by Holton (1996), this thesis focuses on the contribution to trainee motivation of 16 variables identified Holton’s model as:

5

1. secondary influence variables (performance-self efficacy and learner readiness); 2. expected utility variables (transfer effort-performance expectations and performance-outcomes expectations); 3. transfer climate variables (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative and supervisor sanctions); 4. ability variables (personal capacity for transfer and opportunity to use); and 5. enabling variables (content validity and transfer design) The second model used in this thesis is the Theory of Planned Behaviour (TPB) model devised by Ajen (1991) (see Chapter 3) from which the knowledge sharing behaviour variables were tested. Thus, the conceptual framework for this thesis amends the model by Holton (1996) by incorporating the sharing behaviour elements (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) to provide a more expanded view of the transfer of training framework. It should be noted that Holton did not consider knowledge sharing in his HRD model. The development of the research questions for this study arose in response to the paucity of research investigating the question of how transfer of training variables and the variables associated with sharing behaviour perform in the Malaysian public sector. Research questions one and two explore the similarities and differences of transfer of training and sharing behaviour across three training types (general training; management/leadership training; computer training) and demographics (gender; age; level of education; work experience; position of employment) respectively. These two research questions are important because it gives important information on which transfer of training variables are strong, weak or moderate. Having the knowledge on which transfer variables were strong, moderate or poor may assists the researcher to proceed with further analysis to identify which among the strong variables are significant in explaining the variation in motivation to

6

transfer training. This thesis anticipates that there is no significant difference in these variables across training types and demographics and that the variables can be generalised. This issue is taken up in Chapter 3. Research question three was devised to explore which variables explain the variation in motivation to transfer and thus provides information allowing the development of recommendations to HRD managers on how to improve trainees’ motivation to transfer their training. Research questions four to six were devised to explore the possible linkages between sharing behaviour and motivation to transfer, to explore trainees’ intention to share and to explore the direct and indirect relationships (via the significant predictors identified in research question three) between sharing behaviour and motivation to transfer. The six research questions are listed below: Research Question One: Which of these transfer of training variables: •

motivation to transfer;



secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use);



enabling (content validity, transfer design); and



TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean

score

across

different

training

types

(general

training,

management/leadership training, computer training)?

7

Research Question Two: Which of these transfer of training variables: •

motivation to transfer;



secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use),



enabling (content validity, transfer design); and



TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across trainees’ demographics (gender, age, level of education, work experience, position of employment)?

Research Question Three: Which of these transfer of training variables: •

secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use); and



enabling (content validity, transfer design) serve as key significant predictors of one’s motivation to transfer training?

8

Research Question Four: Is the variable: intention to share significantly correlated with sharing behaviour and is sharing behaviour significantly correlated with motivation to transfer? Research Question Five: What are the significant predictors of intention to share? Research Question Six: What are the direct and indirect relationships (via the significant predictors identified in research question three) between sharing behaviour and motivation to transfer? Figure 1.1 depicts the conceptual framework used in this thesis to answer the research questions and Chapter 3 discusses the framework in detail. As discussed briefly above, the conceptual framework was designed to explore the thesis that trainees’ motivation to transfer training is influenced by a number of secondary influences variables, expected utility variables, transfer climate variables, enabling variables and ability variables. The three antecedents to intention to share are also included in the framework along with perceived behavioural control toward knowledge sharing. Table 1.1 provides the definitions of the variables used in this thesis based on those identified by Holton, Bates and Ruona (2000) and Ajzen (1991) respectively.

9

Figure 1.1 The Conceptual Framework

Attitude toward knowledge sharing

Subjective norms toward knowledge sharing

Perceived behavioural control toward knowledge sharing

TPB Variables

Intention to share

Secondary Influences

Personality characteristic variable

Intervention readiness variable

Performanceself efficacy

Learner readiness

Transfer effortperformance expectations Performanceoutcomes expectations

Motivation to transfer

Motivation Elements

Feedback Peer support Supervisor support Openness to change Personal outcomes-positive Personal outcomes-negative Supervisor sanctions

Environmental Elements

Ability/Enabling Elements

Sharing behaviour

Personal capacity for transfer Opportunity to use

Content validity Transfer design

Expected utility variables

Transfer climate variables

Enabling variables

Ability variables

10

Table 1.1 The Definitions of the Variables No

Variables

1

Learner readiness

2

Motivation to transfer

3

Peer support

4

Supervisor support

5

Personal outcomespositive Personal outcomesnegative

6 7

Supervisor sanctions

8

Content validity

9

Transfer design

10

Personal capacity for transfer

11

Opportunity to use

12

15

Performance self efficacy Transfer effortperformance expectations Performance-outcomes expectations Feedback

16

Openness to change

13 14

Definition Extent to which trainees are prepared to enter and participate in training . Trainees’ desire to use the knowledge and skills mastered in the training program on the job. Extent to which peers reinforce and support use of learning to the job. Extent to which supervisors/managers support and reinforce use of training on the job. Degree to which applying training on the job leads to outcomes that is positive for the trainees. Extent to which individuals believe that not applying skills and knowledge learned in training will lead to negative personal outcomes. Extent to which individuals perceive negative responses from supervisors/managers when applying skills learned in training. Extent to which trainees judge training content to accurately reflect job requirements Degree to which (1) training has been designed and delivered to give trainees the ability to transfer learning to the job (2) training instructions match job requirements. Extent to which individuals have the time, energy and mental space in their work lives to make changes required to transfer learning to the job. Extent to which trainees are provided with or obtain resources and tasks on the job enabling them to use training on the job. Trainee’s general belief that they are able to change their performance when they want to. Expectation that effort devoted to transferring learning will lead to changes in job performance. Expectation that changes in job performance will lead to valued outcomes. Formal and informal indicators from an organisation about an individual’s job performance Extent to which prevailing group norms are perceived by trainees’ to resist or discourage the use of skills and knowledge acquired in training.

Source: Holton, EF III., Bates RA & Ruona, WEA 2000, ‘The development of a generalised learning transfer system inventory’, Human Resource Development Quarterly, vol.11, no.4, pp.333-360. 17

Sharing behaviour

18

Intention to share

19

Attitude toward knowledge sharing Subjective norm toward knowledge sharing Perceived behavioural control toward knowledge sharing

20 21

The degree to which trainees actually share the learned knowledge and skills in the workplace. The degree to which trainees are willing to share the learned knowledge and skills in the workplace. Trainees’ positive or negative evaluations on sharing the learned knowledge and skills in the workplace. Perceived social pressure to share the learned knowledge and skills in the workplace. Perceived ease or difficulty of sharing the learned knowledge and skills in the workplace.

Source: Variables 17 to 21 adapted from Ajzen, I 1991, ‘The theory of planned behaviour’, Organisational Behaviour and Human Decision Processes, vol.50, pp.179-221. 11

1.4 Justification for the Research The justification for this thesis is twofold: First, it addresses a key gap in the research and theory on transfer of training through identifying the key variables linked with increasing the likelihood that trainees will transfer their learning onto the job. This is particularly important in a Malaysian context where there is a scarcity of contemporary research in the area. Second, and related to the first, the thesis is justified as it will provide HRD managers with the tools to enhance transfer of training and thus ROI on training course implementation.

1.4.1 Significance of the Thesis The thesis makes a number of contributions in transfer of training research. Notably, it is the first empirical study in the Malaysian public sector to investigate factors which contribute to trainees’ motivation to transfer training. Using the HRD model (Holton 1996), the thesis examines the hypothesised relationships depicted in the research framework (see Figure 1.1). Second, by including knowledge sharing behaviour in the basic Holton (1996) model, it extends both the HRD model and the TBP model (Ajzen 1991) in predicting trainees’ intention to share their learned knowledge and skills in the workplace. Third, the thesis contributes to the design of methodology in transfer of training research through validating the constructs under study using exploratory and confirmatory factor analysis. There is a need for a wellvalidated and generalised instrument to measure transfer of training factors, particularly in the Malaysian public sector where there has been little research. Having a well-validated and generalised instrument to measure the transfer factors is important to ensure confidence in the results obtained. In addition to its academic significance, the findings of the thesis have implications for the practice of HRD. For instance research questions one and two provide HRD managers in the Malaysian public sector with the ability to identify the variables that increase transfer of training behaviours and intervene to reduce those variables that do not enhance transfer. Through the identification of these factors, the thesis also contributes to training evaluation practices which reflect transfer of training as an

12

ROI measure taking into account specific trainee personality characteristics, the transfer climate and transfer design conducive to optimal transfer of training outcomes. Similarly, the findings from research question three provide a better understanding of the factors that motivate trainees to transfer training to the job which can be used by HRD managers to enhance training design and preparation of trainees. Finally, the findings from research questions four, five and six will contribute to HRD managers’ understanding of sharing behaviour as a potential factor in motivating trainees to transfer training to the job. The implications described above are important for HRD practices in the Malaysian public sector to close the gap between what is learned as a result of training and what is applied on the job. This may contribute to greater productivity and ROI in training.

1.5 Research Approach A cross sectional study of the Malaysian public sector was used in this thesis with a survey questionnaire as the main instrument for data collection. The sample chosen consisted of government employees attending training at the National Institute of Public Administration, a central training organisation for government employees in Malaysia. The instrument consisted of an 87 Likert item questionnaire, ranging from ‘1 = Strongly Disagree’ to ‘5 = Strongly Agree’ which was designed to measure the constructs under study. Given the importance of questionnaire design to ensure valid and reliable measurement of all the variables under study, the framework of Churchill (1979), Spector (1992) and Cavana, Delahaye and Sekaran (2001) in questionnaire design were adopted and modified to suit the needs of this thesis. Questionnaire design and administration is discussed in Chapter 3.

1.6 Outline of the Thesis This thesis is arranged into eight Chapters, a bibliography and appendices. This Chapter has described the research background, research problem, and the justification for conducting the research. Chapter 2 commences the literature review with a discussion of the concepts of training, transfer of training and what is meant by

13

motivation to transfer training. The two key training evaluation models used in this study are also detailed and compared to gain greater understanding into the many factors which influence trainees’ motivation to transfer training. The theory of planned behaviour (Ajzen 1991) is also discussed in Chapter 2 which is used in this thesis to examine trainees’ sharing behaviour in the workplace. Chapter 3 develops the conceptual framework based on the two key models by Holton (1996) and Aizen (1991). The Chapter describes the formulation of hypotheses and details the methodology chosen for data collection and analysis. The Chapter concludes with a discussion of the questionnaire design and how it was administered. Chapter 4 continues the methodology section through an assessment of the questionnaire’s construct validity and reliability using exploratory factor analysis, confirmatory factor analysis and Cronbach’s alpha. The Chapter demonstrates that all measures have good psychometric properties (validity and reliability) and are thus appropriate to investigate the relationships hypothesised in this thesis. The testing of the hypotheses belonging to research questions one and two are discussed in Chapter 5 while Chapter 6 discusses the testing of hypotheses for the remaining four research questions. Chapter 5 reveals that transfer of training variables did

not

vary

much

across

the

three

training

types

(general

training;

management/leadership training; computer training) and demographics (gender; age; level of education; work experience; position of employment). The Chapter highlights the weak and strong variables with respect to motivation to transfer and discusses the factors which promote and inhibit transfer of training. Chapter 6 continues with the testing of hypotheses for research question three, four, five and six and answers research questions three to six. The evolution of the hypothesised structural model to the development of the final structural model is detailed, demonstrating that the thesis makes a key contribution in adding sharing behaviour to the traditional HRD models of motivation to transfer training.

14

Chapter 7 discusses the managerial implications, research limitations and suggestions for future research agenda. The Chapter stresses the importance for HRD practices in the Malaysian public sector to diagnose transfer of training variables strategically and intervene to prevent the operation of variables that inhibit transfer. The Chapter provides advice to HRD managers at two broad levels: pre-training level and posttraining to assist them enhance motivation to transfer. Finally, Chapter 8 concludes this study with a summary of the main points from each chapter and lists the major observations arising from the empirical research and the surrounding literature.

1.7 Research Limitations and Assumptions Whilst this thesis makes a significant contribution to research, theory and HRD application of transfer of training factors, it is acknowledged that there are several limitations of the study and that a number of assumptions had to be made along the way in order to operationalise the research questions. These limitations necessarily affect the generalisability of the findings and are discussed below: First, the data were collected from a purposive sampling of government employees attending training at the National Institute of Public Administration, Malaysia in the month of August 2005 until September 2005. This means that trainees in this study represent only a sub-sample of all trainees attending training in the year 2005. Thus, the findings should be generalised with some caution even within the Malaysian public sector. Second, this study uses self-reporting for all the variables under investigations. It is acknowledged that results of a single source of data may be affected by method variance (Podsakoff & Organ 1986). For example, motivation to transfer training is based solely on trainees’ perceptions but not assessed by their supervisors. Obtaining data from supervisors on what can motivate trainees to transfer their training could increase confidence in the results. It is an area for future research but was outside of the practical limits of the present work.

15

Third, the number of variables used when modelling with Structural Equation Modelling was kept as a manageable set due to the small sample size obtained in this study. The researcher recognised that there are other potential variables such as learning, opportunity to use and transfer design that may have an impact on motivation to transfer training and these, again may be used to broaden the research framework in future research. Fourth, although the scales used in this study fell within the desired range of validity and reliability, they were not tested against a different set of data due to the difficulty in obtaining a large number of respondents at this time of study for this purpose. This was largely due to a number of training programs being cancelled due to low participation and accommodation problems. Using a different set of data for construct reliability and validity may have increased further the confidence in the results obtained. Finally, for modelling with SEM in this study, the final structural model was revised to improve fit. Again, this means that the findings reported in this thesis should only be generalised with caution.

Apart from the limitations listed above, this thesis proceeded on the basis of two key assumptions. First, the items in the questionnaire were held to cover all the variables under investigation. This assumption was checked and validated by two panel experts in Malaysia but nevertheless, given the size of the study, it is open to speculation that further factors may be elucidated. Second, the study relied on the assumption that the respondents answered the questionnaire truthfully.

1.8 Summary This Chapter laid the foundations for the thesis by noting the significant gap in transfer of training research in the Malaysian public sector and the lack of adequate evaluation to ensure that the training was being transferred back into the workplace.

16

The Chapter indicated that this study situates itself in this research and HRD knowledge gap with its objectives to identify more closely, the factors which contribute to transfer of training, motivation to transfer and to test whether knowledge sharing behaviour plays a role in the transfer of training regime. The Chapter then moved to a discussion of the justifications for the research, both academic and managerial, concluding that significant contributions are expected through its modification of the key academic transfer models and through the recommendations for HRD practice which may enhance return on investment for the training effort. The next Chapter traces the key international literature on transfer of training, motivation to transfer training and knowledge sharing behaviour in the workplace.

17

CHAPTER 2 TRANSFER OF TRAINING: A REVIEW OF THE RESEARCH LITERATURE

2.1 Introduction Chapter 1 outlined the background for this thesis indicating that transfer of training research has been sparse in Malaysia despite increasing pressure on the public sector to deliver a return on investment in terms of improved training transfer and the productivity gains likely to accompany this. The Chapter described preliminary evidence suggesting that motivation is a factor which enhances training transfer. Chapter 2 aims to set the research problem in its academic context by canvassing the international research literature on transfer of training before moving to a detailed discussion of the concept of motivation to transfer. First, the Chapter outlines the various definitions of training and transfer of training in order to set a base line for understanding the operational variables in this area. The Chapter then sets out the definitions of motivation to transfer training and the third section explores the factors which may influence trainees’ motivations to transfer their training by considering the evolution of several key training evaluation models: the Kirkpatrick (1994) four level evaluation model and the Learning Transfer System Inventory (LTSI) (Holton et al. 2000). The Human Resource Development Evaluation Research and Measurement Model (the HRD model) (Holton 1996) is described in the fourth section of this Chapter. The HRD model was the first attempt to comprehensively specify factors that can influence trainee’s motivation to transfer training both directly and indirectly. As discussed in Chapter 1, the HRD model failed to include the concept of knowledge sharing behaviour as a variable which influences transfer of training. This is addressed in the final section of this chapter which draws together a discussion of knowledge sharing behaviour and describes its benefits. It is

18

hypothesised in this thesis that knowledge sharing behaviour plays a role in facilitating transfer of training. For this reason, TPB theory (Ajzen 1991) is discussed within a framework that explains a trainee’s intention to share the knowledge and skills learned in training with others in the workplace

2.2 Training and Transfer of Training The International Encyclopaedia of Adult Education and Training (1996:519) defined training as the provision that is aimed at creating intentional learning processes that contribute to improving the performance of workers in their present job. The definition does not differ significantly from definitions of training in the HRD context. For instance, in an HRD environment, training is often defined as a planned learning experience designed to bring about permanent change in an individual.’s knowledge, attitudes, or skills (Campbell, Dunnete, Lawler & Weick 1970:497). Goldstein (1992:3) provided a definition that related training to individual performance which is, arguably, a more apt descriptor of HRD objectives. He defined training as the systematic acquisition of attitudes, concepts, knowledge, roles or skills that result in improved performance at work. Generally, it has been found that most workplace training definitions in the international literature emphasise the current job as the focus. For instance, Tziner, Haccoun and Kadish (1991) noted that the fundamental purpose of training is to help people develop skills and abilities which, when applied at work, will enhance their average job performance in their current job. The definition provided by Tziner et al. (1991) links the acquisition of knowledge and skills gained through training to an application in the workplace. This link represents the concept of training transfer. Transfer of training is generally defined as the degree to which trainees apply the knowledge, skills and attitudes gained in training to their job (Ford & Weissbein 1997; Tannenbaum & Yulk 1992; Wexley & Latham 1991). Most researchers used the terms ‘transfer of training’ and ‘transfer of learning’ interchangeably to refer to the application of the knowledge and skills learned in training back to the job. The

19

application of these skills has also been described as an ongoing exercise rather than a once-off task. In this sense, transfer of training has been described as the maintenance of skills, knowledge and attitudes over a certain period of time (Baldwin & Ford 1988). Transfer of training needs to be considered as a multidimensional construct because different authors view transfer of training differently, attributing a variety of features to its definition. For example, Wexley and Latham (1991) suggest that transfer can be measured as a positive, negative or a zero. Positive transfer occurs when learning in the training situation results in better performance on the job. This reflects the general assumption behind most definitions of transfer of training. Negative transfer occurs when learning in the training situation results in poorer performance on the job. Zero transfer, not surprisingly, occurs when learning in the training situation has no effect on the job performance. Other researchers have provided different insights into transfer of training. For example, Cormier and Hagman (1987) considered it to consist of two elements: general or specific transfer. On this view, general transfer refers to the application of learned knowledge and skills to a higher level or to a more complex work situation. It occurs when a trainee has grasped the generic skills or concepts and generalised their application (for instance, problem solving). Specific transfer occurs when the trainee can apply what has been learned in the training environment to a similar work situation (for instance, learning to use a word processor in training with application of that learning at work). Finally, Laker (1990) proposed a distinction between near transfer and far transfer in a training context. According to this author, near transfer occurs when trainees apply what was acquired in training to situations very similar to those in which they were trained. Far transfer, in contrast, occurs when trainees apply the training to different situations from the ones in which they were trained.

20

Regardless how the transfer of training elements have been described, there has been general agreement amongst researchers that transfer of training is a critical issue in HRD. For instance, Baldwin and Ford (1988), in their early model of the transfer process provided HRD researchers and practitioners of organisational training with an understanding of the range of factors affecting transfer of training include a range of trainee characteristics, the training course design and the type of work environment. Further, many researchers in this area have emphasised that any effort taken to evaluate training effectiveness must look for these elements of transfer of training (Broad & Newstrom 1992; Kirkpatrick 1994; Noe 2005; Noe et al. 2004). According to Bates (2003), training can do little to increase individual or organisational performance unless what is learned as a result of training is transferred to the job. The gap between what is learned and what is applied on the job represents, at least in HRD terms, a massive transfer problem (Baldwin & Ford 1988; Broad & Newstrom 1992; Ford 1994). In one study, Broad and Newstrom (1992) surveyed 85 trainees and asked them how much of the material learned was used on the job over time. The responses were: immediately – 41 percent; six months later - 24 percent; and one year later – 15 percent. Broad and Newstrom (1992) also noted that the lack of involvement of line managers and the lack of reinforcement on the job were major barriers to the transfer of training. Not surprisingly, it has been reported in the literature that a mere of 10 percent of the investment in training is returned in performance improvement (Garavaglia 1993; Georgenson 1982). Despite the reported problems in achieving effective transfer of training reported in the international HRD research, the training and development of employees continues to be viewed as a key strategy for organisations to gain a competitive advantage (Goldstein 1992; Noe 2005; Wexley & Latham 1991). One factor which stands out in the literature as a contributor to more effective transfer has been the extent to which trainees are motivated to use their training on the job. The next section considers the research on motivation to transfer training.

21

2.3 Motivation to Transfer Training Many researchers have acknowledged that transfer of training will occur only when trainees have the motivation or desire to use the learned knowledge and skills on the job (Baldwin & Ford 1988; Noe 1986; Noe & Schmitt 1986; Wexley & Latham 1991). Arguably, without motivation to transfer, even the most systematic training program will struggle to be effective. However, little is known about the specific factors that impact on a trainee’s motivation to transfer training to the job (Seyler et. al. 1998; Tannenbaum & Yulk 1992). As this thesis is concerned with uncovering factors that could influence a trainee’s motivation to transfer his or her training to the job, it is pertinent to examine the two key training evaluation models in the HRD literature. The Kirkpatrick (1994) training evaluation model (see 2.3.1 below) and the Learning Transfer System Inventory (see 2.3.2 below) (Holton et al. 2000) have received the most attention by researchers in the area of training evaluation. Other training evaluation models such as developed by Hamblin (1974), Phillips (1995) and (Brinkerhoff 1987) are not discussed because they are encompassed in the extended version of Kirkpatrick’s (1994) model. For example, Hamblin (1974) included economic benefits (in a five-level model) while Brinkerhoff (1987) proposed a six-level model. Phillips (1995) focused on return on investment in his model. Although each of these other authors’ work contributes greatly to the knowledge in training evaluation, their models largely mirror Kirkpatrick’s model and it is to this model the thesis now turns.

2.3.1 The Kirkpatrick Model The Kirkpatrick (1994) model of training evaluation, also known as the four-level evaluation model, has dominated the field of training evaluation for more than 30 years (Alliger & Janak 1989; Alliger, Tannenbaum, Bennet, Traver & Shortland 1997). As depicted in Figure 2.1 below, the model consists of four stages: reaction→learning→behaviour →results. Kirkpatrick (1994) described reaction, learning, behaviour and results as training outcomes (measures that organisations use

22

to evaluate training programs) that an intervention (the training course) hopes to create. Figure 2.1 The Kirkpatrick Four Level Evaluation Model

Level 1

Reaction

Level 2

Learning

Level 3

Behaviour

Level 4

Results

Source: Adapted from Kirkpatrick, DL 1994, Evaluating training programs. The Four Levels, Berrett-Koehler Publishers, San Francisco.

Based on the model, reaction is the first level in the evaluation process and is defined as how well the trainees were satisfied with a particular training program (Kirkpatrick 1994:27). According to Kirkpatrick, evaluating reaction is important for several reasons. First, it will give valuable feedback and suggestions for improving future training programs. Second, it tells trainees that the trainers are there to help them do their job better and that they need feedback to determine how effective they are. Third, reaction sheets can provide quantitative information to managers and those who concerned about the program as well as to establish standards of performance for future training programs. In order for trainers to get maximum benefit from reaction measures, Kirkpatrick (1994:28) provided guidelines for evaluating reaction that included: designing a form that will quantify the reaction to training; encouraging written comments and suggestions which can be useful in the redesign of the training course; aiming for a 100 percent response; developing acceptable standards (for instance, standards for instructors or facilities) and to measure reactions against standards.

23

Next, learning was defined as the extent to which trainees change their attitudes, improve their knowledge and increase their skills as a result of attending training (Kirkpatrick 1994:42). Kirkpatrick stressed that evaluating learning is important because no change in behaviour (level 3) can be expected unless learning objectives have been accomplished. The suggested guidelines for learning included using a control group (if practical); using a paper-and-pencil test to evaluate the learned knowledge and skills both before and after the program and using the evaluation results to take appropriate action, particularly for trainers to work on being more effective in teaching in the future (Kirkpatrick 1994:43). Behaviour is the third level in the evaluation process and is defined as the extent to which change in behaviour has occurred because the trainee attended the training program (Kirkpatrick 1994:52). This level attempts to measure how much transfer of knowledge, skills, and attitudes occurs. Kirkpatrick stressed that it is important to see whether the knowledge and skills learned in the training program were transferred to the job. However, he warned that no evaluation should be attempted until trainees had an opportunity to use the new learning. Thus, for some training programs, two or three months after training may be appropriate. Evaluating learning behaviour can be done by using surveys or through interviewing one (or more) of the trainees, their immediate supervisor, their subordinates and others who are knowledgeable about their behaviour (Kikpatrick 1994:55). Finally, level four of the Kirkpatrick model is the evaluation of results. Results can be defined as what occurred as a consequence of the trainees attending the training program (Kirkpatrick 1994:63). Kirkpatrick used this fourth level to relate the results of the training program to organisational objectives (for instance, increased production, improved quality, decreased costs, increased sales and higher profits). He stressed that the final objectives of the training program need to be stated according to the organisational objectives for optimal return on investment. The suggested guidelines to evaluate results included using a control group (to eliminate factors other than training that could have caused the changes observed); measuring both

24

before and after the training program and considering cost versus benefits (the value of the actual results compared to the cost of the training program). According to Kirkpatrick (1994) there is a natural flow between the levels in the model. Reaction could lead to learning; learning could lead to behaviour change; and change in behaviour could lead to positive organisational results. As the model depicted only four variables (reaction, learning, behaviour and results) the effect of trainees’ motivation to transfer training to the job was not an overt consideration of Kirkpatrick’s. However, it is likely that motivation was an assumption underpinning the model as Kirkpatrick (1994:51) wrote that ‘without learning, no change in behaviour will occur’. This corresponds broadly with research which has demonstrated that changes in behaviour occur when trainees are motivated to use their training on the job (Holton 1996; Noe & Schmitt 1986; Seyler et. al. 1998). The findings from previous studies have also shown that other variables lying outside the training classroom also affect behaviour change (Baldwin, Magjuka & Loher 1991; Hicks & Klimoski 1987). These external factors have been identified as the transfer climate (Holton, Bates, Seyler & Carvalho 1997; Rouiller & Goldstein 1993), workplace design (Kupritz 2002) and personality characteristic variables such as selfefficacy (Gist 1989; Gist, Schwoerer & Rosen 1989; Gist, Steven & Bavetta 1991) and readiness to participate in training could also affect behaviour change. Despite its dominance in the field, as described above, the Kirkpatrick (1994) model did not provide a strong guide to understanding what influences trainees’ motivation to use training. In light of more recent literature on motivation, this could be a key omission in this traditional model. It is relevant then to turn to the other dominant training evaluation model in the HRD literature: the Learning Transfer System Inventory (LTSI) developed by Holton, Bates and Ruona (2000).

25

2.3.2 The Learning Transfer System Inventory (LTSI) According to Holton, Bates and Ruona (2000), organisations wishing to enhance the ROI from training need to understand the factors that affect transfer of training and then intervene to remove the factors which inhibit transfer. Indeed, the authors argued that the first step to improving transfer is to accurately diagnose the inhibiting factors. In their 2000 study titled “Development of a Generalised Learning Transfer System Inventory”, they introduced the concept of transfer system which encompassed factors in the person, training and organisation that influence transfer of learning to job performance. Figure 2.2 The LTSI Conceptual Evaluation Model

Secondary Influences

Motivation Elements

Outcomes

Learning

Individual Performance

Environmental Elements

Organisational Results

Ability/Enabling Elements

Source: Adapted from Holton, EF III 1996, ‘The Flawed Four-Level Evaluation Model,’ Human Resource Development Quarterly, vol.7, no.1, p.9.

The LTSI conceptual evaluation model used to develop the LTSI is depicted in Figure 2.2. Three primary training outcomes were defined in this model. These outcomes were: learning; individual performance; and organisational results. Learning was defined as: achievement of the learning outcomes desired in an intervention; change in individual performance as a result of the learning being

26

applied on the job; and results at the organisational level as a consequence of the change in individual performance (Holton 1996:9). In comparison with Kirkpatrick’s (1994) training evaluation model, three primary differences are of note. First, there is an absence of reaction as a training outcome in the LTSI. Holton, Bates and Ruona (2000) argued that reactions should be removed from evaluation models citing several studies which indicated that reactions had no significant relationship with learning (Alliger and Janak 1989; Dixon 1990; Noe & Schmitt 1986; Warr & Bunce 1995). For example, Warr and Bunce (1995) divided reactions into three components (enjoyment, usefulness and perceived difficulty) and they found no significant correlation between any of them and learning outcomes. Second, individual performance is used instead of behaviour in the Kirkpatrick (1994) model because Holton et al. (2000) claimed that individual performance is a broader construct than behaviour change and a more appropriate descriptor of HRD objectives. And third, the LTSI conceptual model is arguably a more comprehensive model than the Kirkpatrick (1994) model because it accounts for the impact of motivation, environmental and ability/enabling elements. The evolution of the LTSI into its present form occurred across the late 1990s. Following the development of Holton’s (1996) conceptual evaluation model (Figure 2.2), a study by Holton, Bates, Seyler and Carvalho’s (1997) identified a number of climate variables which affect transfer of training. In this 1997 study, the authors found that trainees perceived transfer climate according to referents to their organisation (for example supervisor, peer/task, or self) and the factor analysis in this study extracted seven transfer climate constructs. These constructs are detailed in Table 2.1 and listed below: •

supervisor support (a supervisor’s reinforcement of use of training on the job); opportunity to use (trainees are provided with resources enabling them to use training on the job);



peer support (peers support use of learning to the job);

27



supervisor sanctions (perception of negative responses from supervisors when applying skills learned in training);



personal outcomes-positive (applying training on the job leads to positive outcomes for trainees);



personal outcomes-negative (applying training on the job leads to negative outcomes for trainees); and



openness to change (prevailing group norms discourage the use of skills and knowledge acquired in training)

Further, factor analysis from the data also suggested two further transfer design constructs: content validity (trainees judge training content to accurately reflect job requirements) and transfer design (training has been designed to provide ability to transfer learning to the job and instructions match job requirements). The authors then searched the literature on transfer of training to identify seven other constructs that had not been previously tested in Holton et al.’s (1997) study but which they believed, would fit into the conceptual model. The seven additional variables are detailed in Table 2.1 and comprise performance-self efficacy (the belief that trainees are able to change their performance when they want to) (Gist 1989; Gist et al. 1989; Gist et al. 1991); two expectancy related variables: transfer effort-performance expectations and performance-outcomes expectations (expectation that effort devoted to transferring learning will lead to changes in job performance and outcomes respectively) (Bates & Holton 1999; Noe & Schmitt 1986); personal capacity for transfer (trainees make the changes required to transfer learning to the job) (Ford, Quinones, Sego & Sorra 1992); feedback (formal and informal indicators about an individual’s job performance); learner readiness (trainees prepared to participate in training) (Knowles, Holton & Swanson 1998); and motivation to transfer (trainees’ desire to use the knowledge and skills mastered in the training program on the job) (Noe 1986; Noe & Schmitt 1986). Figure 2.3 shows how the sixteen variables fit into the conceptual model and Table 2.1 provides definitions for each of the 16 final transfer variables.

28

Figure 2.3 The Learning Transfer System Inventory (LTSI)

Secondary Influences

Motivation Elements Environmental Elements

Performance self-efficacy Learner readiness Motivation to transfer Transfer —> Performance Performance —> Outcomes Feedback Peer support Supervisor support Openness to change Learning

Personal outcomes-positive Personal outcomes-negative Supervisor sanctions

Individual Performance

Organisational Results

Outcomes Ability

Content validity Transfer design Personal capacity for transfer Opportunity to use

Source: Holton, EF III, Bates, RA & Ruona WEA 2000, ‘The development of a generalised learning transfer system inventory,’ Human Resource Development Quarterly, vol.11, no.4, p.339.

Figure 2.3 shows how the LTSI accounts for the impact of primary variables such as environmental, ability and motivational variables. The LTSI indicates that motivation to transfer is influenced by secondary variables such as performance-self efficacy and learner readiness. Holton et al. (2000) refer to the variables affecting an individual’s performance in the LTSI as a transfer system, which they defined as all the factors in the person, training and organisation that influence transfer of learning to job performance. In other words, they argued that transfer must occur before learning can lead to individual job performance. Table 2.1 below details the definitions of the variables as depicted in the LTSI.

29

Table 2.1: The 16 factors of the LTSI which affect transfer of training No

Variables

1

Learner Readiness

2 3

Motivation to Transfer Peer Support

4

Supervisor Support

5

Personal Outcomespositive Personal Outcomesnegative

6 7

Supervisor Sanctions

8

Content Validity

9

Transfer Design

10

Personal Capacity to Transfer

11

Opportunity To Use

12

15

Performance Self Efficacy Transfer EffortPerformance Expectations PerformanceOutcomes Expectations Feedback

16

Openness to Change

13 14

Definition Extent to which trainees are prepared to enter and participate in training. Trainees’ desire to use the knowledge and skills mastered in the training program on the job. Extent to which peers reinforce and support use of learning to the job. Extent to which supervisors/managers support and reinforce use of training on the job. Degree to which applying training on the job leads to outcomes that is positive for the trainees. Extent to which individuals believe that not applying skills and knowledge learned in training will lead to negative personal outcomes. Extent to which individuals perceive negative responses from supervisors/managers when applying skills learned in training. Extent to which trainees judge training content to accurately reflect job requirements Degree to which (1) training has been designed and delivered to give trainees the ability to transfer learning to the job (2) training instructions match job requirements. Extent to which individuals have the time, energy and mental space in their work lives to make changes required to transfer learning to the job. Extent to which trainees are provided with or obtain resources and tasks on the job enabling them to use training on the job. Trainee’s general belief that they are able to change their performance when they want to. Expectation that effort devoted to transferring learning will lead to changes in job performance. Expectation that changes in job performance will lead to valued outcomes. Formal and informal indicators from an organisation about an individual’s job performance Extent to which prevailing group norms are perceived by trainees’ to resist or discourage the use of skills and knowledge acquired in training.

Source: Holton, EF III, Bates, RA & Ruona WEA 2000, ‘The development of a generalised learning transfer system inventory,’ Human Resource Development Quarterly, vol.11, no.4, pp.344-346.

30

Several studies have successfully used the LTSI model to validate the factors affecting transfer of training (Chen 2003; Donovan et al. 2001; Holton et al. 2003; Yamnill 2001). For example, Chen (2003) found that the LTSI was valid in Taiwan and Yamnill (2001) validated it in Thailand. The LTSI model has also been claimed to be influential in measuring training effectiveness (Donovan et. al. 2001). Although the LTSI model included motivation to transfer as one of the variables that could affect individual performance, the model only specified two secondary influence variables (learner readiness and performance-self efficacy) that could influence motivation to transfer. In order to gain an in-depth understanding of the direct and indirect effect on motivation to transfer, this chapter now moves to discuss the Human Resource Development Evaluation Research and Measurement Model (the HRD model) (Holton 1996). This model, according to Holton (1996) should be used for research purposes in investigating motivation to transfer training.

2.4 The Human Resource Development (HRD) Evaluation Research and Measurement Model The HRD model was developed by Holton (1996) and is the first attempt to comprehensively specify the antecedent relationships leading to motivation to transfer. As depicted in Figure 2.4, the model shows all hypothesised relationships indicated by thick arrows (primary relationships) or lighter arrows (secondary relationships). Primary intervening variables (ability, motivation to learn, reaction to learning, transfer design, motivation to transfer, transfer climate, expected utility, linkage to organisational objectives and external events) are shown in boxes with arrows pointing directly to one of the outcomes. Secondary intervening variables (intervention readiness, job attitudes, personality characteristics, and intervention fulfilment) are linked by arrows in Figure 2.4 to the primary intervening variables. The primary intervening variables are hypothesised to influence the outcomes (learning; individual performance; and organisational results) that an intervention is targeted to achieve whereas the secondary intervening variables are hypothesised to 31

have a secondary influence on the motivation elements (motivation to learn; motivation to transfer). Reaction is included in the model and is hypothesised to have a moderating role in the relationship between motivation to learn and learning. Two studies were found using this model investigating factors affecting motivation to transfer training (Bates & Holton 1999; Seyler et al. 1998). Figure 2.4 The HRD Evaluation Research and Measurement Model

Intervention Secondary Readiness Influences Personality Job Characteristics Attitudes

Motivation Elements Environmental Elements

Outcomes Ability/enabling elements

Motivation to Learn

Motivation to Transfer Transfer Climate

Reaction

Learning

Intervention Fulfillment

Individual Performance

Ability

Transfer Design

Expected Utility/ROI External Events

Organisational Results

Linkage to Organisational Goals

Source: Holton, EF III 1996, ‘The Flawed Four-Level Evaluation Model’, Human Resource Development Quarterly, vol.7, no.1, p.17.

This section overviewed the two key training evaluation models used in HRD research: the Kirkpatrck (1994) model and the LTSI model, which evolved during the late 1990s from the early conceptual framework published by Holton in 1996. The main differences between the models were highlighted, particularly in light of emerging research suggesting that motivation to transfer training is now increasingly regarded as an important criteria for transfer of training (Holton 1996; Noe & Schmitt 1986; Seyler et. al. 1998). Finally, the HRD model (Holton 1996) was described as an

32

attempt to comprehensively specify the antecedent relationships leading to motivation to transfer. The next section moves to a discussion of the motivational elements in Holton’s (1996) HRD model in light of the broader HRD research literature. This section is divided into four parts. The first discusses the influences on motivation to transfer, the second discusses the influences on motivation to learn, the third addresses the influences on learning outcomes and finally, the influences on individual performance are considered. The influences on motivation to learn, learning outcomes, and individual performance are also discussed because they are important to gain insight on their contribution to motivation to transfer. The influences on organisational results such as external events (for example, equipment failures, raw material shortages) and linkage to organisational goals (interventions that are closely linked to organisational goals) are excluded from this discussion because whilst they provide insight into the success or failure of organisational training, they are not related to motivation to transfer based on Holton’s (1996) model.

2.4.1 Influences on Motivation to Transfer In Holton’s (1996) model, five categories of variables are hypothesised to influence motivation to transfer: job attitudes, intervention fulfilment, expected utility, transfer climate and learning outcomes. They form some of the transfer variables used in this thesis and are detailed below.

Job Attitudes Job attitudes refer to trainees’ attitudes toward the organisation and the job (Holton 1996:11). According to Noe and Schmitt (1986), highly job-involved individuals are more likely to be motivated to learn new skills because participation in training activities can increase skill levels and improve job performance. Several studies have investigated the relationship between job attitudes and motivation to transfer but their findings have been mixed (Cheng & Ho 2001; Clark 1990; Kontoghiorghes 2004;

33

Mathieu, Tannenbaum & Salas 1992). For example, Mathieu et al. (1992) did not find a significant relationship between job involvement and motivation to transfer for the reason that trainees who were highly involved in their jobs did not see the training program as instrumental in obtaining valued outcomes. A similar finding was found in Cheng and Ho’s (2001) study this time for the reason that trainees pursuing a postgraduate program may represent their desires to enhance their employability rather than their job performance. On the other hand, two studies demonstrated the significant effect of the variables: job involvement on motivation to transfer (Clark 1990; Kontoghiorghes 2004). Despite the slightly mixed findings on job attitudes, it remains a likely element with an influence on motivation to transfer training. In this thesis, job attitudes was not included as a variable under investigation because it is categorised in Holton’s (1996) model as a secondary influence variable that affects motivation to transfer rather than being a primary influence. According to Holton (1996), it may be sufficient to measure a few secondary influence variables where much of the variance can be explained. In this chapter, those secondary influence variables that were examined in this thesis will be highlighted.

Intervention Fulfilment Intervention fulfilment refers to the extent to which training meets or fulfills the trainee’s expectations and desires (Holton 1996:13). The effect of intervention fulfilment on motivation to transfer training has received little attention. Only one study was found to test this notion and the finding has supported the relationship as hypothesised in Holton’s (1996) model (Tannenbaum, Mathieu, Salas, & CannonBowers 1991). In Tannenbaum et al.’s (1991) study, the relationship between intervention fulfilment and motivation to transfer was significant. Therefore, they suggested that for intervention fulfilment to be a useful concept in understanding trainees’ motivation to transfer, future research should assess intervention fulfilment in other training environments. In this thesis, intervention fulfilment was not examined because this variable is categorised as secondary rather than primary influence variable that affects motivation to transfer.

34

Expected Utility According to Holton (1996:13), training programs that have expected utility or exhibit a payoff to both the organisation and to trainees should result in greater motivation to transfer learning to the job. This notion is consistent with Vroom’s (1964) expectancy theory that individuals will be more motivated when they believe their effort invested in the training program will result in mastery of the training content (effort-performance expectation) and when they believe that good performance in the training program will lead to desirable outcomes (performance-outcome expectation). Two studies examined the relationship between expected utility and motivation to transfer (Bates & Holton 1999; Clark, Dobbins & Ladd 1993). Bates and Holton (1999) studied the role of expectancies in the motivation to transfer in a social service agency. In this study, motivation to transfer was conceptualised firstly as a function of utility (or expectancy beliefs) about the extent to which learning is expected to have useful job application; and secondly, of rewards or the extent to which the application of learning on the job is perceived to result in some valued outcome for the individual. Findings from this study suggested that utility was a significant predictor of motivation to transfer while rewards were not. This finding has some resonance with training in the Malaysian public sector (indeed, likely all public sector organisations) where extrinsic rewards are often not available to motivate employees (Hameed & Analoui 1999) and where employees have been found to respond to more intrinsic rewards (Poon & Idris 1985). For example, in Hameed and Analoui’s (1999) study, the researchers found that the lack of intrinsic rewards in the form of recognition by employers inhibited trainees from putting into practice what they learned in training. In another study, Poon and Idris (1985) investigated employees’ perceptions and attitudes towards how their rewards are and should be determined. The researchers found that intrinsic rewards in the forms of interesting, meaningful and challenging work and feedback on performance were found to be highly valued. Other studies have also demonstrated the value of expectancy theory (Vroom 1964) in explaining motivation to learn and training transfer (Noe & Schmitt 1986; Yamnill

35

2001; Chen 2003). For example, Noe and Schmitt (1986) found that expectations regarding effort-performance and performance-outcome linkages were highly correlated with motivation to learn. In this thesis, expected utility was examined as a variable affecting motivation to transfer because it is categorised as a primary variable in Holton’s (1996) model.

Transfer Climate Transfer climate is hypothesised in Holton’s (1996) model as an important environmental element to influence motivation to transfer. According to Holton (1996), trainees who worked in conditions supportive of training transfer are more likely to transfer their learning to the job. Seyler et al. (1998) conceptualised transfer climate to refer to organisational climate that includes supervisor support, supervisor sanctions and peer support. This study demonstrated that peer support and supervisor sanctions were significant predictors of motivation to transfer but supervisor support was not for the reason that little unique variance was left to be explained by supervisor support after accounting for the influence of other organisational climate variables. Although supervisor support was not significant in Seyler et al.’s (1998) study, other studies have demonstrated its positive effect (Clarke 2002; Clark et al. 1993; Kontoghiorghes 2001) and that of peer support (Clark et al. 1993; Ruona, Leimbach, Holton & Bates 2002) on motivation to transfer. Transfer climate was also conceptualised to consist of situations and consequences that either inhibit or help to facilitate the transfer of learning into a job situation (Rouiller & Goldstein 1993). These authors suggested four types of ‘situational’ cues: goal cues, social cues, task cues and self-control cues operate in the transfer process. These situational cues remind trainees of what they have learned, or at least provide an opportunity for them to use what they have learned. In contrast, ‘consequence’ cues were described as on-the-job outcomes, which affect the extent to which training is transferred. The four ‘consequence’ cues comprise positive feedback, negative feedback, punishment, and no feedback. Unfortunately, their study could not validate the suggested constructs because the sample size was inadequate. Later, Holton et al.

36

(1997) continued the effort to validate the transfer climate variables suggested by Rouiller and Goldstein (1993). Results of the later study suggested that trainees perceive the transfer climate according to referents to the organisation (for example supervisor, peer/task, or self) rather than according to psychological cues (for example goal cues, social cues), as proposed by Rouiller and Goldstein (1993). Besides peer support, supervisor support and supervisor sanctions, the study identified other transfer climate constructs including openness to change (prevailing group norms are perceived to discourage use of new skills); personal outcomespositive (application of training on the job leads to positive outcomes); personal outcomes-negative (application of training on the job leads to negative outcomes); and opportunity to use learning (trainees are provided with resources and tasks that enable them to use their new skills on the job). Another study by the Holton team attempted to validate transfer climate constructs in an effort to develop a diagnostic tool to measure factors affecting transfer of training (Holton et al. 2000). In this study, feedback (formal and informal indicators from an organisation about an individual’s job performance) emerged as yet another dimension of transfer climate. A range of studies have confirmed the effect of feedback on training transfer (Clarke 2002; Tracey, Tannenbaum & Kavanagh 1995) and employees’ performance (Reber & Wallin 1984). For example, Clarke (2002) found, in a study of training in the Social Services Department, UK, that an absence of feedback to the trainee on his or her performance impedes the transfer of training. The study confirmed earlier research by Reber and Wallin (1984) who predicted that the performance of employees will be enhanced if they received feedback about their own goads when related to their department’s performance. Research examining personal outcomes-positive on training outcomes was found in two studies (Bates & Holton 1999; Tracey et al. 1995). For example, in Tracey et al.’s study (1995) found that extrinsic personal outcomes (for example, pay and promotion) and intrinsic personal outcomes (for example, praise and recognition) have a direct impact on post training behaviours. In particular, extrinsic rewards

37

exhibited very weak associations with training retention whereas intrinsic rewards proved to be more important variable positively impacting on training retention. Two other studies examined personal outcomes-negative (Kontoghiorghes 2001; Tracey et al. 1995). Tracey et al. (1995) found that personal outcomes-negative such as punishment was a significant predictor of post training behaviour, particularly in inhibiting training retention. A contrary finding was found in Kontoghiorghes’s (2001) study when punishment for failing to use the new skills exhibited only weak associations with training retention. Despite the contradictory finding in Kontoghiorghes’s (2001) study, this thesis examines transfer climate as a variable affecting motivation to transfer because it is categorised as a primary variable in Holton’s (1996) model and this has been confirmed in other studies showing the significant influence of transfer climate on training transfer and motivation to transfer.

Learning Outcomes Several studies have examined the relationship between learning outcomes and motivation to transfer (Huczynski & Lewis 1980; Seyler et al. 1998; Tannenbaum et al. 1991). In Tannenbaum et al.’s (1991) study, learning was assessed using test performance following training. The authors hypothesised that test performance was related to motivation to transfer. Indeed, they found that test performance was positively related to motivation to transfer. In a similar study, Huczynski and Lewis (1980) also reported that motivation to transfer was influenced by the learning gained. An interesting, contrary finding to these two studies was that reported by Seyler et al. (1998) who found that learning was not a significant predictor of motivation to transfer. The authors reasoned that the lack of findings related to learning may be a function of the way learning was measured. In their particular study, learning was measured by averaging test scores recorded by a computer on tests taken by the trainees at the end of each lesson. The authors reported that they were not given the opportunity to audit the tests and therefore, there was no assurance that the tests were representative measures of learning that took place during training.

38

In Holton’s model, learning was categorised as a primary variable affecting motivation to transfer. However, in this thesis, learning was not examined because the researcher did not have the opportunity to meet with the trainers to discuss how learning was measured.

2.4.2 Influences on Motivation to Learn The previous section detailed the five variables in Holton’s (1996) HRD Evaluation Research and Measurement Model which influence motivation to transfer. This section now turns to the influences on a trainee’s motivation to learn, which in turn, has been linked with increased transfer of training. Motivation to learn can be defined as the specific desire of a learner to learn the content of a training program (Noe 1986; Noe & Schmitt 1986). In Holton’s (1996) model, motivation to learn is hypothesised to be influenced by personality characteristics, intervention readiness, and job attitudes. These elements are the source of variables used in this study and will now be discussed.

Personality Characteristics Baldwin and Ford (1988), in their model of the transfer process, suggested selfefficacy as one of the trainee’s characteristics that can affect training transfer. Selfefficacy is a key element in Bandura’s Social Learning Theory refering to the belief in one’s own capability to perform a specific task (Bandura 1977). The concept of selfefficacy can be understood from three dimensions: magnitude or level, strength and generality. Magnitude or level applies to the level of task difficulty. In other words, people may differ in their self belief of being capable of performing tasks of varying difficulty. Strength, the second dimension of self-efficacy, refers to whether the conviction regarding magnitude is strong or weak. This means that individuals may differ in their confidence in attaining a given level of performance. Finally, generality indicates the degree to which the expectation is generalised across situations (Bandura 1977:194).

39

Psychologists have contributed greatly in the field of human HRD, especially in understanding of how self-efficacy affects the application of knowledge, skills and behaviour learned in training on the job. This is evidenced by researchers who have found that self-efficacy has been shown to be related to motivation to learn (Colquitt, LePine & Noe 2000; Quinones 1995) and training outcomes: training and task performances (Gist 1989; Gist et al. 1989; Gist et al. 1991; Tannenbaum et al. 1991). For example, in Gist’s (1989) study, self-efficacy was found to be positively related to training performance on an innovative problem solving task. Another study by Gist et al. (1989) also found self-efficacy played an important role in computer software training when trainees with high levels of self-efficacy performed better than trainees with lower levels. This finding was consistent with a later study by Gist et al. (1991) that examined the effects of self-efficacy in a two-stage training process on the acquisition and maintenance (retention) of complex interpersonal skills (negotiation skills, in this instance). The authors found that trainees with high self-efficacy negotiated significantly higher salaries than trainees with moderate or low selfefficacy. These studies have demonstrated that the effectiveness of training was dependent on the strength of trainees’ self-efficacy. Whilst many studies had attempted to examine the effect of self efficacy on training transfer, no study has been located to examine the effect of self efficacy on motivation to transfer. For this reason, in this thesis, self-efficacy was examined as a secondary influence variable affecting motivation to transfer because it is likely that trainees with high self-efficacy are motivated to transfer training as well.

Intervention Readiness Intervention readiness is hypothesised in the model to influence motivation to learn. According to Holton (1996) intervention readiness includes such variables as the degree to which trainees are involved in assessing training needs; involved in planning the training; the degree to which their expectations are clarified; the degree of choice; and other unexplored influences. Research examining trainees’ readiness to participate in training was found in four studies (Baldwin et al. 1991; Hicks & Klimoski 1987; Ryman & Biersner 1975; Tannenbaum et al. 1991). In Hicks and

40

Klimoski’s (1987) study, intervention readiness was conceptualised as referring to prior information that trainees receive about their training program and the amount of freedom they have to take the program. The authors found that trainees who received a realistic training preview (containing some positive, neutral and unfavourable statements) and those who had a high degree of choice were more likely to believe the training program was appropriate for them to take and were better able to benefit from the training. These trainees also showed more commitment to their decision to attend training than others who received the traditional announcement (containing brief overly positive statements) and those who had a low degree of choice. In a similar study, Baldwin et al. (1991) examined the degree of choice of training content rather than simply the choice to attend training in general as was done in Hicks and Klimoski’s (1987) study. Baldwin et al. (1991) found that trainees who received their choice had a higher level of motivation to learn prior to entering the training session than those who were not provided the choice. In this thesis, intervention readiness was examined as a secondary influence variable affecting motivation to transfer because it is likely that trainees who are ready to participate in training are motivated to transfer training as well.

Job Attitudes As described earlier, job attitudes are said influence motivation to transfer. In addition, job attitudes are also hypothesised to influence motivation to learn as well. Four studies have examined the relationship between job attitudes and motivation to learn but findings are mixed (Cheng & Ho 2001; Facteau, Dobbins, Russell, Ladd & Kudisch 1995; Mathieu et al. 1992; Noe & Schmitt 1986). For example, Noe and Schmitt (1986) found that job involvement was strongly related to the acquisition of the key behaviours emphasised in the training program while in Cheng and Ho’s (2001) study, job involvement was not significantly related to learning motivation and learning transfer for the reason that the trainees tested (who were pursuing an MBA degree) may have represented their desires to enhance their employability rather than job performance. In this thesis, job attitudes was not examined as a

41

variable affecting motivation to transfer and the justification for this has been described previously in section 2.4.1.

2.4.3 Influences on Learning Outcomes In the third part of this discussion on elements affecting motivation, this chapter turns to the influences on learning outcomes. Specifically, learning outcomes are hypothesised in Holton’s (1996) model as being influenced by ability, reaction to training and motivation to learn. These variables will now be discussed.

Ability Ability refers to the general capacities related to performance of a set of tasks (Fleishman 1972). Psychologists have demonstrated that general cognitive ability has been shown to have substantial positive relationship with job performance (Ford et al. 1992; Kanfer & Ackerman 1989; Robertson & Downs 1979). For example, Robertson and Downs (1979) in their review of trainability studies (the degree to which trainees are able to learn and apply the material emphasised in the training program) have suggested that approximately 16 percent of the variance in trainee performance may be attributable to ability. Research on ability was also done to see how it interacts with motivation to enhance outcomes. For example, Kanfer and Ackerman (1989) found that individual differences in cognitive abilities clearly exerted an effect on performance. In another study, Ford et al. (1992) found that those trainees with high abilities obtained a higher number of the trained tasks when performed on the job. In this thesis, ability was examined as a variable affecting motivation to transfer because it is likely that trainees with high ability are more motivated to transfer training to the job.

Trainee Reactions As described in section 2.3.1 above, Kirkpatrick’s (1994) four level training evaluation

model

(reaction→learning→behaviour

change→outcome)

defined

reaction as trainees’ ‘liking of’ and ‘feelings for’ a training program. In his model, he

42

viewed reaction as an outcome that could lead to learning. However, he also stressed that a favourable reaction to a training program does not assure learning. Several studies examined the relationship between reaction and learning and the findings showed no support for this relationship. For example, a meta analytic review of the literature based on Kirkpatrick’s model conducted by Alliger and Janak (1989) found that only eight studies reported the correlation between reaction and learning and the correlation reported was weak. Therefore, they concluded that reactions had no significant relationship with learning. Similarly, Noe and Schmitt (1986) also found no support for a direct link between reaction and learning while in Dixon’s (1990) study, reaction was found to have a little correlation with learning. In Holton’s (1996) model, reaction was not viewed as an outcome but as having a moderating role between motivation to learn and learning. Two studies were located to test this notion but the findings were mixed. Support for the hypothesised relationship was found in Mathieu et al.’s (1992) study. In their study, reaction was found as having a moderating role between motivation to learn and learning as well as acting as a mediator of other relationships. However, a contrary finding was found in Seyler et al.’s (1998) study but the authors gave reason that it was due to learning measurement problems. The authors explained that although the learning measure was based on tests created by subject matter experts, there was no assurance that the tests were comprehensive or representative measures of the learning took place during the training. In this thesis, reaction was not examined because it has no direct effect on motivation to transfer (Holton 1996). Further, as described earlier, several other studies have shown the non significant effect of reaction on learning (Alliger & Janak 1989; Noe & Schmitt 1986).

Motivation to Learn Motivation to learn can be defined as the specific desire of a learner to learn the content of a training program (Noe 1986; Noe & Schmitt 1986). Motivation to learn

43

is hypothesised in Holton’s (1996) model to positively influence learning outcomes. Two studies examined this relationship but the findings were mixed. For example, Quinones (1995) found that whilst motivation to learn had a positive relationship with learning and behaviour it was not related to task performance. Contrary to the findings of Quinones (1995), Noe & Schmitt (1986) in their model of motivational influences on training effectiveness, found no significant relationship between motivation to learn and learning. In this thesis, motivation to learn was not examined because despite being linked to learning outcomes in Holton’s (1996) study it demonstrated no direct effect on motivation to transfer in the Holton model.

2.4.4 Influences on Individual Performance The final set of variables linked to motivation to transfer is that of individual performance. Individual performance is hypothesised in Holton’s (1996) model to be influenced by motivation to transfer, learning, transfer design and transfer climate. Research examining the relationship between motivation to transfer and individual performance has been largely neglected but an argument can be sustained for their inclusion in this thesis on the basis that indirect evidence has linked them to motivation to transfer. For instance, learning has been shown to be correlated with motivation to transfer (Huczynski & Lewis 1980; Tannenbaum et al. 1991) and behaviour change (Alliger et al. 1997). Therefore, one can argue that an individual’s performance is more likely to occur in highly motivated and successful learners. Further, research has also demonstrated that transfer design (Bates, Holton, Seyler & Carvalho 2000; Yamnill 2000) and transfer climate (Ford et al. 1992; Seyler et al. 1998) were significant predictors of motivation to transfer, which in turn, affect individual performance. The previous section considered in detail, the motivational elements in Holton’s (1996) HRD Evaluation Research and Measurement Model along with a range of studies which also tested these variables. First, the influences on motivation to transfer were considered and detailed, particularly with relevance to the set of variables used for this thesis. Second, the discussion moved to the influences on 44

motivation to learn. A number of variables: personality characteristics, intervention readiness, and job attitudes were detailed and a rationale was provided for the inclusion of the first two as variables used in this thesis. Third, the influences on learning outcomes were considered. Of the variables discussed: ability, reaction to training and motivation to learn the thesis utilised ability as one variable to be tested. Finally, the influences on individual performance were briefly explored with a view to including it as a variable in this thesis. These influences together provide an important insight into motivation to transfer and they form part of the conceptual framework for this thesis which is described in full in Chapter 3. The next section draws together a discussion on sharing behaviour as another element which affects motivation to transfer training on the job. Whilst this thesis has found Holton’s (1996) model to be both relevant and useful in terms of its insight into the factors that influence motivation to transfer, it is argued that the Holton (1996) model ignored the role that sharing behaviour plays as an indicator of motivation to transfer training. Such a role can certainly be implied from the findings of research in the area. Several transfer researchers have suggested the importance of knowledge sharing in understanding transfer of training (Noe & Ford 1992; Roscow & Zager 1988; Tracey et al. 1995). However, knowledge sharing behaviour has not been fully explored in the transfer literature. Only one study was located which examined this variable and it was found that knowledge sharing behaviour did have an impact on post-training behaviour by influencing self-efficacy and motivation (Tracey et al. 1995). Given that this provides prima facie evidence that knowledge sharing behaviour acts to promote motivation to transfer, the inclusion of this variable in the thesis forms a key hypothesis (this is discussed further in Chapter 3). The argument for including knowledge sharing behaviour as a factor impacting on motivation to transfer training in the Malaysian public sector is discussed below.

45

2.5 Knowledge Sharing and Its Benefits Connelly and Kelloway (2003:294) defined knowledge sharing as a set of behaviours that involves the exchange of information or assistance to others. One proposed definition consists of donating and collecting knowledge, where knowledge donating refers to communicating to others what one’s personal intellectual capital is, and knowledge collecting refers to consulting colleagues in order to get them to share their intellectual capital (Van Den Hoof & Ridder 2004:118). It has been recognised that knowledge is valuable to organisations, particularly when it is shared (Noe et. al. 2004). As more organisations move towards achieving the status of a ‘learning organisation’, the sharing of knowledge among employees becomes crucial because it is the key and indeed, defining element in a learning organisation (Gephatt, Marsick, Van Buren & Spiro 1996). A learning organisation is a company that has an enhanced capacity to learn, adapt and change (Gephatt et al.1996). In a learning organisation, the people are the essential ingredients and therefore, in order to become a learning organisation, the people must be committed to learning and willing to share what they have learned (Noe et. al. 2004). There has been a growing interest in the impact of knowledge sharing in a training and development context as a mechanism to meet business challenges and provide competitive advantage (Noe 2005; Noe et. al. 2004). According to Noe (2005:433), employees are expected to acquire new skills and knowledge in training, apply them on the job and share this information with fellow workers. This has led the changing role of training from a focus on programs to a broader focus on learning and creating and sharing knowledge (Martocchio & Baldwin 1997; Noe 2005; Noe et. al. 2004). The changing of training’s role according to Noe (2005) is depicted in Figure 2.5 below.

46

Figure 2.5 Evolution of Training’s Role

Training Event

Performance Result Create and Share Knowledge

Business Need

Learning Emphasis

Source: Noe, RA 2005, Employee Training and Development, p.41.

Figure 2.5 demonstrates that to some extent, training will continue to focus on developing programs to teach specific skills. However, Noe (2005) argued, to improve employees’ performance and help meet business needs and challenges means that the role of training must necessarily evolve to emphasise learning, creating and sharing knowledge. The importance of knowledge sharing behaviour in understanding transfer of training has been acknowledged by several researchers through the concept of continuous learning (Noe & Ford 1992; Rosow & Zager 1988; Tracey, Tannenbaum & Kavanagh 1995). However, only one study appears to have empirically examined the relationship between continuous learning and post-training behaviour (Tracey et. al. 1995). In Tracey et al.’s (1995) study, continuous learning was conceptualised as the acquisition, application and sharing of knowledge, behaviours and skills not only from training but also from a variety of other sources. The findings of that study indicated that continuous learning (often conceptualised as knowledge sharing) has a positive effect on post-training behaviour. In other words, the more trainees share 47

their knowledge, the greater the transfer of training. Tracey et al. (1995) called for future research to examine how knowledge sharing may affect post-training behaviour by influencing self-efficacy, motivation and trainees expectations about training experiences (Tracey et al. 1995). Clearly, it follows then, that if sharing behaviour does have a direct effect on self-efficacy and motivation, then trainees in a less supportive transfer climate will probably be less likely to transfer training to the job. Although sharing behaviour has not been fully explored in the training transfer literature, the benefits of knowledge sharing have been documented in other settings. For example, knowledge sharing has been found to: promote better learning by individuals (Collison & Cook 2004), provide an important mechanism to achieve better decision making (Tschannen-Moran 2001); and could lead to project effectiveness (Eisenhardt & Tabrizi 1995; Henderson & Cockburn 1994; LeanordBarton & Sinha 1993). Further, the benefits of knowledge sharing have been reported in studies of firms such as Buckman Laboratories and Texas Instruments, which claimed significant gains in revenue (Chua 2003) while Dow Chemicals and Chevron reported savings (Stewart 2001). One important point raised by researchers and practitioners in this field is that the success of knowledge sharing strategies in organisations is dependent on the extent to which individuals are willing to share their knowledge (Connelly & Kelloway 2003; Lin & Lee 2004; Sveiby & Simons 2002; Van Den Hoof & De Ridder 2004). Without the willingness to share, the benefits of knowledge sharing described earlier are less achievable. Therefore, it may be illuminating to examine the attitude towards knowledge sharing behaviour in order to gain a greater understanding of why individuals decide to engage or not to engage in knowledge sharing. This issue is taken up in the following section which details the Theory of Planned Behaviour (Ajzen 1991), a theory which has its roots in the field of social psychology and which has been widely used to predict and explain the phenomena of behavioural intention and actual behaviour.

48

2.6 The Theory of Planned Behaviour The Theory of Planned Behaviour (TPB) originated in the field of social psychology as a predictor for human behaviour (Ajzen 1991; Ajzen & Fishbein 1980; Fishbein & Ajzen 1975). The TPB predicts that the most important determinant of a person’s behaviour is behaviour intent. The individual’s intention to perform a behaviour is a combination of his or her attitude toward performing the behaviour, the prevailing subjective norms and the perceived behavioural controls on the individual (Ajzen 1991). Figure 2.6 outlines the TPB model, which illustrates that one’s intentions give rise to one’s behaviour. Acting on this relationship between intention and behaviour are: one’s attitude towards the behaviour; the subjective norms surrounding the behaviour; and one’s perceived behavioural controls. Based on TPB, peoples’ attitudes towards their own behaviours refer to the degree to which they have made favourable or unfavourable evaluations of the behaviour in question. Subjective norms are the perceived social pressures from significant others to perform or not to perform the behaviour and perceived behavioural controls refer to the perceived ease or difficulty of performing the behaviour (Ajzen 1991:188). In combination, one’s attitude toward the behaviour, subjective norms and perceived behavioural controls lead to the formation of a behavioural intention. According to Ajzen (1991), the more favourable the attitude and subjective norm with respect to the behaviour, and the greater the perceived behavioural control, the stronger should be an individual’s intention to perform the behaviour under consideration.

49

Figure 2.6 The Theory of Planned Behaviour

Attitude toward the behaviour

Subjective norm

Intention

Behaviour

Perceived behavioural control

Source: Ajzen, I 1991, ‘The theory of planned behaviour,’ Organisational Behaviour and Human Decision Processes, vol.50, p.182.

A review of 185 independent studies that used TPB to predict human behaviour by Armitage and Conner (2001) found that, the TPB accounted for 27 to 39 percent of the variance in behaviour and intention respectively, providing support for the efficacy of the TPB as a predictor of intention and behaviour. In the international literature, the TPB has been widely used in empirical research to predict human behaviours in different settings. For example, the theory has been used to better understand consumers’ intentions to attend a sport event (i.e., hockey) (Cunningham & Kwon 2003). In their study attitude, subjective norms and perceived behavioural controls were positively associated with intentions to attend a sport event. Other studies variously utilised TPB to predict individual’s intentions to use a search engine as a learning tool (Liaw 2004); to predict drivers’ compliance with speed limits (Elliot, Armitage & Baughan 2003); to predict hunting behaviours (Hrubes & Ajzen 2001); to predict dishonest actions (Beck & Ajzen 1991); and to predict teachers’ intentions to provide dietary counselling (Astrom & Mwangsi 2000). Further, the

50

TPB has also been applied in a workplace context to assess the extent to which senior managers intended to encourage knowledge sharing (Lin & Lee 2004). Together, these studies point to the efficacy of the TPB in understanding human behaviour through people’s intentions. In terms of the present thesis, the TPB has the potential to provide an insight into an individual’s intention to share the knowledge and skills learned in training with others in the workplace.

2.7 Summary This chapter described the concept of training, transfer of training and motivation to transfer training through an international search of the research literature. While training was defined as a planned learning experience designed to bring about permanent change in an individual’s knowledge, attitudes, or skills, transfer of training was described as the degree to which trainees apply the knowledge, skills and attitudes to their job. A trainee’s motivation to transfer, on the other hand, was described as a trainee’s desire to use the training on the job. Although, over time, researchers have expressed different views about transfer of training by proposing variously: concepts of positive, negative and zero transfer; general and specific transfer; and far and near transfer, it was argued in this chapter, that there was general agreement that transfer of training will occur only when trainees have the desire (motivation) to use the knowledge and skills learned in training on the job. The two dominant evaluation models found in the literature were discussed in order to paint a picture of the factors which may influence a trainee’s motivation to transfer training: the Kirkpatrick (1994) evaluation model and the LTSI (Holton et. al. 2000). The Kirkpatrick (1994) evaluation model provided a starting point in HRD evaluation but ultimately did not provide a strong guide to the whole training process as it focused only on training outcomes (reaction, learning, behaviour and results) and did not account for the impact of other variables such as motivation and work environment on behaviour change. The LTSI (Holton et. al. 2000), on the other hand, provided some indication that motivation to transfer is influenced by secondary influence variables, that is, performance-self efficacy and learner readiness. 51

The chapter then moved to the key model outlining motivation to transfer described the Human Resource Development (HRD) Evaluation Research and Measurement Model (Holton 1996). Based on this model, motivation to transfer is hypothesised to be influenced by intervention fulfilment, job attitudes, learning, expected utility and transfer climate. Other variables such as intervention readiness, personality characteristics, ability, motivation to learn and transfer design were hypothesised to have an indirect influence on motivation to transfer. These factors were supported by several studies discussed in the chapter. Arguably, a key omission in the models presented in this chapter was their failure to include knowledge sharing as a potential variable influencing motivation to transfer training. This chapter introduced the concept of knowledge sharing and reported its benefits. The chapter indicated that the theory of planned behaviour may be used to provide an insight into an individual’s intention to share learned knowledge and skills with others in the workplace, providing a set of variables to test the relationship of knowledge sharing with motivation to transfer training. In the next chapter, the development of conceptual framework and the methodology chosen for the present thesis is detailed.

52

CHAPTER 3 THE CONCEPTUAL FRAMEWORK AND METHODOLOGY

3.1 Introduction Chapter 2 set out the research context to this study by introducing the international literature pertaining to motivation to transfer training. The definitions of motivation to transfer training were considered before moving to a discussion on the various factors known to influence trainees’ motivations to transfer their training. This was done through an examination of the key training evaluation models: the Kirkpatrick (1994) model, the LTSI model (Holton et al. 2000) and the HRD model (Holton 1996). It was also hypothesised in Chapter 2 that knowledge sharing plays a role in facilitating transfer of training and consequently, the TPB theory (Ajzen 1991) was discussed within a framework relating to trainees’ intention to share their knowledge and skills with others in the workplace Chapter 3 now moves to lay out the conceptual framework and methodology used in this thesis to link knowledge sharing with motivation to transfer training. The Chapter describes the hypotheses formulated as the basis of inquiry for the thesis. Finally, the chapter describes the methodology chosen to test the relationships hypothesised.

3.2 The Conceptual Framework As described above and detailed in Chapter 2, the research framework for this study was constructed from an adaptation of three key HRD models: the LTSI model (Holton et al. 2000); the HRD model (Holton 1996) and the TPB theory (Ajzen 1991). First, the category of variables receiving most attention in the literature on the

53

basis of their ability to influence motivation to transfer training was selected for the framework. These variables explain a major portion of the variance in the concept of motivation to transfer training. Added to this were: two secondary variables (personality characteristics and intervention readiness); and four primary variables (expected utility, transfer climate, ability and transfer design). The variables from the Learning Transfer System Inventory (LTSI) developed by Holton et al. (2000) were also fitted into the research framework. As discussed in Chapter 2 the LTSI model was developed from Holton’s earlier 1996 model and as they pertained to motivation to transfer, they fitted well into the conceptual framework. The LTSI variables comprised: secondary variables (performance-self efficacy and learner readiness); expected utility variables (transfer effort-performance expectations and performanceoutcome expectations); transfer climate variables (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomesnegative and supervisor sanctions); ability variables (personal capacity for transfer and opportunity to use); and enabling variables (content validity and transfer design). The definitions for each of these variables is provided in Chapter 1 (see Table 1.1). Finally, the variables from the theory of planned behaviour (TPB) (Ajzen 1991) were included in the framework. As explained in Chapter 2, the TPB describes the elements pertaining to knowledge sharing which is hypothesised here as being linked to motivation to transfer training. Figure 3.1 depicts the conceptual framework developed for this thesis.

54

Figure 3.1 The Conceptual Framework

Attitude toward knowledge sharing

Subjective norms toward knowledge sharing

Perceived behavioural control toward knowledge sharing

TPB Variables

Intention to share

Secondary Influences

Personality characteristic variable

Intervention readiness variable

Performanceself efficacy

Learner readiness

Transfer effortperformance expectations Performanceoutcomes expectations

Motivation to transfer

Motivation Elements

Feedback Peer support Supervisor support Openness to change Personal outcomes-positive Personal outcomes-negative Supervisor sanctions

Environmental Elements

Ability/Enabling Elements

Sharing behaviour

Personal capacity for transfer Opportunity to use

Content validity Transfer design

Expected utility variables

Transfer climate variables

Enabling variables

Ability variables

55

The contribution of this conceptual framework to the understanding of factors influencing motivation to transfer training is fourfold. First, the conceptual framework is unique in its utilisation of the application of TPB to predict trainees’ sharing behaviour in the workplace. As explained in Chapter 2, the theory of planned behaviour predicts that a trainee’s intention to share his or her knowledge and skills in the workplace will be determined by his or her attitude toward sharing behaviour together with the operation of subjective norms and perceived behavioural control. The more favourable the attitude and subjective norms and the greater the perceived behavioural control, the stronger should be trainees’ intention to share the learned knowledge and skills in the workplace. Second, the conceptual framework is unique in its hypothesis that the personality characteristic variable: performance-self efficacy and the intervention readiness variable: learner readiness have a direct influence on motivation to transfer. In contrast, Holton’s (1996) model hypothesised that the personality characteristic and intervention readiness variables influenced motivation to learn. As described in Chapter 2, research has suggested that the personality characteristic variable, self efficacy (Gist 1989; Gist et al. 1989; Gist et al. 1991; Tannenbaum et al. 1991) and intervention readiness variable, learner readiness (Hicks & Klimoski 1987; Baldwin et al. 1991; Tannenbaum et al. 1991; Ryman & Biersner 1975) are related to two key training outcomes: training and task performance. It is from the work of these researchers that this thesis argues that trainees with high self-efficacy and are ready to participate in training are motivated to transfer training. Third, the conceptual framework is unique in its hypothesis that the ability variables: personal capacity for transfer and opportunity to use have a direct influence on motivation to transfer. In Holton’s (1996) model, the ability variables were hypothesised to influence motivation to transfer indirectly via their relationship with learning. Nevertheless, the ability variable, personal capacity for transfer was described as an important determinant of training transfer (Holton et al. 2000; Holton et al. 2003). The variable, opportunity to use was described as a significant predictor

56

of motivation to transfer (Seyler et al. 1998) and associated positively with training transfer (Awoniyi et al. 2002; Lim & Johnson 2002; Tracey et al. 1995). Therefore, it is argued in this thesis that trainees with high personal capacity for transfer and have the opportunity to use their training are more motivated to transfer that training in the workplace. Finally, the conceptual framework hypothesises that the enabling variables, content validity and transfer design exert a direct influence on motivation to transfer. In Holton’s (1996) model, transfer design was hypothesised to influence individual performance. Content validity was found to have a significant correlation with motivation to transfer (Seyler et al. 1998) and was a significant predictor of transfer performance (Bates et al. 2000; Axtell, Maitlis & Yearta 1997). Earlier work on transfer design had found that using identical elements (for example when training environment was identical to the work environment) (Gagne, Baker & Foster 1950), teaching through general principles (for instance, trainees taught not just applicable skills but also the general rules and theoretical principles that underlie the training content) (Bernstein, Hillix & Marx 1957) and using several examples of a concept to be learned (Shoe & Sechrest 1961) resulted in training transfer. Therefore, given the reported influence of the enabling variables, content validity and transfer design on motivation to transfer and actual transfer of training, it was hypothesised that content validity and transfer design exert a direct influence of on motivation to transfer. As described in Chapter 2, a number of factors included in the original models used to derive the conceptual framework [the Kirkpatrick (1994) model, the LTSI model (Holton et al. 2000), the HRD model (Holton 1996) and the TPB (Ajzen 1991)] were removed. The resulting conceptual framework removed: reaction, learning, external events, organisational performance and linkage to organisational goals. The reasons for their removal from the conceptual framework is described in Chapter 2 and summarised in Table 3.1:

57

Table 3.1

Factors removed from the original HRD models

FACTOR Reaction

REASON FOR REMOVAL This variable is not significantly correlated with learning (Noe & Schmitt 1986; Alliger & Janak 1989; Dixon 1990) and is not found to moderate the relationship between motivation to learn and learning (Seyler et al. 1998). Further, reaction was not hypothesised in Holton (1996) model to have an influence on motivation to transfer.

Learning

The researcher did not have the opportunity to examine whether the material used for the performance test during training were representative measures of the learning that took place during training.

external events,

These factors are not related with motivation to transfer training

organisational

(Holton 1996).

performance and linkage to organisational goals

3.2.1 The Research Questions Based on the conceptual framework, this thesis attempts to answer the six specific research questions: Research Question One: Which of these transfer of training variables: •

motivation to transfer;



secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);

58



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use);



enabling (content validity, transfer design); and



TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean

score

across

different

training

types

(general

training,

management/leadership training, computer training)? Research Question Two: Which of these transfer of training variables: •

motivation to transfer;



secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use),



enabling (content validity, transfer design); and



TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across trainees’ demographics (gender, age, level of education, work experience, position of employment)?

59

Research Question Three: Which of these transfer of training variables: •

secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use); and



enabling (content validity, transfer design) serve as key significant predictors of one’s motivation to transfer training?

Research Question Four: Is the variable: intention to share significantly correlated with sharing behaviour and is sharing behaviour significantly correlated with motivation to transfer? Research Question Five: What are the significant predictors of intention to share? Research Question Six: What are the direct and indirect relationships (via the significant predictors identified in research question three) between sharing behaviour and motivation to transfer? In order to answer the above research questions, this thesis formulated a series of hypotheses (H1 to H10) and they are stated in Table 3.2:

60

Table 3.2 The Statement of Hypotheses 1

Hypothesis (H) H1: These transfer of training variables: motivation to transfer; secondary influences (performance-self efficacy, learner readiness); expected utility (transfer effort-performance expectations, performance-outcomes expectations); transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomesnegative, supervisor sanctions); ability (personal capacity for transfer, opportunity to use), enabling (content validity, transfer design) and TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across different training types (general training, management/leadership training, computer training).

2

H2: These transfer of training variables: motivation to transfer; secondary influences (performance-self efficacy, learner readiness); expected utility (transfer effort-performance expectations and performance-outcomes expectations); transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions); ability (personal capacity for transfer and opportunity to use), enabling (content validity and transfer design) and TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing and perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across trainees’ demographics (gender, age, level of education, work experience, position of employment).

3

H3: Secondary influences variables (performance-self efficacy, learner readiness) will explain a significant proportion of variance in motivation to transfer. H4: Expected utility variables (transfer effort-performance expectations, performanceoutcomes expectations) will explain a significant proportion of variance in motivation to transfer. H5: Transfer climate variables (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions) will explain a significant proportion of variance in motivation to transfer. H6: Enabling variables (content validity, transfer design) will explain a significant proportion of variance in motivation to transfer. H7: Ability variables (personal capacity for transfer, opportunity to use) will explain a significant proportion of variance in motivation to transfer.

4

H8: Intention to share will be significantly correlated to sharing behaviour and sharing behaviour will be significantly correlated to motivation to transfer.

5

H9: Attitude, subjective norm and perceived behavioural control toward knowledge sharing will explain a significant proportion of variance in intention to share.

6

H10: Sharing behaviour will have a direct and indirect relationship (via the significant predictors identified in research question three) with motivation to transfer.

61

This section described the conceptual framework developed for this study. The framework was derived from four key HRD models with respect to their contribution to understanding the concept of one’s motivation to transfer training. From the conceptual framework, six research questions and 10 hypotheses were presented as the key areas of inquiry for this study. The next section describes the methodology chosen to test the 10 hypotheses formulated in this thesis in order to answer the six research questions.

3.3 Methodology This section chronicles the methodology utilised for the thesis commencing with a description of the questionnaire design, the sample chosen and the procedures undertaken for data collection. The chapter then moves to consider how data screening was conducted, how the checking of multivariate assumptions was undertaken and how construct validity and reliability were examined. The final part of this section describes the statistical techniques used for hypothesis testing.

3.3.1 Questionnaire Design The variables depicted in the conceptual framework were measured using multiple items in the questionnaire. For this reason, the researcher searched the literature to find validated scales for the 21 constructs. However, it was found that only sample of items were reported (normally one item) in the journals and some of the scales were copyrighted (Holton et al. 2000). Thus, the researcher developed the scales to measure the 21 constructs and the leading methodologists in scale development were consulted (Cavana et al. 2001; Churchill 1979; De Vellis 2003; Hinkin 1995; Spector 1992). The survey instrument was developed in Bahasa Malaysia (Malay Languange) and the English version was included in Appendix A for reporting purposes. It comprised of a 87 Likert item questionnaire designed to measure the constructs under study. The questionnaire utilised a five-point scale that ranged from ‘1=Strongly Disagree’ to

62

‘5=Strongly Agree’. Questionnaire design followed a framework of nine steps which is described below. The Framework of Questionnaire Design The framework used to develop the questionnaire was based on Churchill (1979:66), Spector (1992:8) and Cavana et al. (2001:228). Churchill’s (1979) framework was originally developed for marketing research but it has been applied to other disciplines as well such as for developing a measure of knowledge management behaviours and practices (Darroch 2003); for developing a measure of participative decision making (Parnell & Bell 1994); and for developing a measure of online learning (Fortune, Shifflett & Sibley 2006). Spector’s (1992) framework was developed purposely for summated rating scales (multiple item scales) and therefore, was considered appropriate for this thesis. Finally, Cavana et al.’s (2001) framework was used in this study because it takes into account the principle of wording and the general appearance of the questionnaire. The three frameworks were modified to the needs of this thesis. The modified framework consists of nine steps as depicted in Figure 3.2 and detailed below. Step 1-Define Construct The first step in the questionnaire design was to define the constructs of interest (Spector 1992; Churchill 1979). Churchill (1979:67) stressed that researchers must be exacting in delineating what is included in the definition and what is excluded to ensure that what we want to measure is determined clearly. In this thesis, 16 constructs were defined based on the LTSI model (Holton et al. 2000) and another five constructs were based on the TPB model by Ajzen (1991) (see also Chapter 2). For example, one of the constructs based on the LTSI model (Holton et al. 2000) is learner readiness which refers to the extent to which trainees are prepared to enter and participate in training. Therefore, the definition provided by Holton et al. (2000) was then used to measure learner readiness. Similarly, all other constructs under study were given the definition prescribed in the originating model from which they were drawn.

63

Figure 3.2 The Framework for Developing Questionnaire D e fin e co n stru ct

S te p 1

D e te rm in e re sp o n se ch o ice s

S te p 2

G e n e ra te sa m p le o f item s

S te p 3

D e te rm in e qu e stio n se qu e n ce

S te p 4

D e te rm in e la yo u t a n d a p pe a ra n ce

S te p 5

E xp e rt ju dge m e n t a n d revise d

S te p 6

P ilo t stu d y

S te p 7

Ite m a n a lysis a n d re vise d

S te p 8

F in alisin g qu e stion n a ire

S te p 9

(Source: Churchill 1979; Cavana et al. 2001; Spector 1992)

64

Step 2-Determine Response Choices The second step in the process of questionnaire design was to determine the nature of responses available to respondents. The three most common response choices are agreement, evaluation and frequency (Spector 1992:19). Agreement asks respondents to indicate the extent to which they agree with each item. Evaluation asks for a rating for each item based on aptness of response. Frequency asks for a judgement of how often items have occurred, should, or usually occur. Most studies in transfer of training using questionnaires as the instrument in data collection apply a five-point Likert-type scale to indicate the extent to which respondents agree or disagree with each item, as well as measure the magnitude of agreement or disagreement (for example, Holton et al. 2000; Seyler et al. 1998; Yamnill 2000; Chen 2003). Although in some studies, a seven-point (for example, Machin & Fogarty 2004) and a 10-point (for example, Gaudine & Saks 2004) scale has been used, there is a body of research suggesting that reliability increases as the number of points increase to five and then level off as the points increase to five (Lissitz & Green 1975). Therefore, this study utilised a five point scale ranging from 1 (Strongly Disagree) to 5 (Strongly Disagree). Step 3-Generate Sample of Items The third step in the framework of questionnaire design was to generate a sample of items for all constructs under study. In order to guide this step, several recommendations made by researchers were taken into consideration as follows: • The content of each item should primarily reflect the construct of interest (De Vellis 2003:63). • A large number of items represents a form of insurance against poor internal consistency (De Vellis 2003:66). • Lengthy items should be avoided as length usually increases complexity and diminishes clarity (De Vellis 2003:67; Cavana et al. 2001:232). • A measure should have at least three relatively homogeneous items for content adequacy (Cook, Hepworth, Wall & Warr 1989:4).

65

• Items that convey two or more ideas (doubled-barrelled) should be avoided (De Vellis 2003:68; Churchill 1979:68; Spector 1992:23) • Negatively worded items should be used to avoid agreement bias (De Vellis 2003:69; Churchill 1979:68; Spector 1992:24). • Double negative worded items are a source of ambiguity (Baker 2003:352). • Information that respondents cannot or will not provide should not be asked for (Schwab 2005:43). • It should be made simple (Schwab 2005:43). • It should be specific (Schwab 2005:43). • It should use plain language (Spector 1992:25). • It should use everyday language, simpler words and simple sentences (Baker 2003:351). • It should consider the reading level of respondents (Spector 1992:25). • It should consider the inclusion of validation items (De Vellis 2003:87). In addition to the recommendations above, the researcher conducted a focus group interview with four subjects (officers at the Ministry of Finance), who had recently experienced a training course. Using focus group interviews during the item generation stage was suggested by Churchill (1979) and employed by other transfer of training researchers (for example, Enos, Kehrhahn & Bell 2003; Hayes & Pulparampil 2001). Specifically, the objective of the focus group interview was to find out what trainees understood by each of the 21 concepts under study. The interview was tape-recorded with their permission and transcribed. The themes emerging from the transcribed interview were then used for generating items (Appendix B lists the items generated for each construct). During the item generation process, the researcher did not conduct any sorting procedures involving subject matter experts. The involvement of HRD experts were in Step 6 where they were invited to examine each item and make a judgement on whether each item does measure the theoretical construct nominated (see Step 6, pg.67).

66

Step 4-Determine Question Sequence The fourth step in the questionnaire design was to determine the item sequence. In order to guide this step, the researcher adopted several recommendations made in other studies: •

Use a funnel approach. This means, each item in the questionnaire should be determined from the general to the specific and from items that are relatively easy to answer to those that are progressively more difficult (Cavana et al. 2001:232).



Negatively and positively worded items should be placed in different parts of the questionnaire, as far apart as possible (Cavana et al. 2001:232).



Place the items randomly to reduce any systematic biases in the response (Cavana et al. 2001:232).



Relegate sensitive items to the body of the questionnaire and intermix among some not-so-sensitive ones (Churchill 1979:205).

Step 5-Determine Layout and Appearance The layout and general appearance of the questionnaire are important to ensure that the questionnaire looks attractive (Cavana et al. 2001:234). In this step, the researcher prepared the covering page containing information about the researcher such as the name, thesis title, the objectives of the study and inviting respondents to participate. Other information such as the total number of questions to be answered, the time needed to complete the questionnaire, contact information and most importantly, the statements about the confidentiality and anonymous of the information provided were also stated (see Appendix A). Other criteria such as the selection of font size (12, Times New Roman) and line spacing (1.5) were also taken so that the questionnaire appeared neat and attractive to enhance questionnaire completion by the target group. Step 6-Expert Judgement and Revised Upon the completion of Step Five, the questionnaire was examined for content validity. According to Cavana et al. (2001:238), assessing content validity can be

67

done by a group of experts who examine each item and make a judgement on whether each item does measure the theoretical construct nominated. For this purpose, the questionnaire was sent to two academics in the area: one from Universiti Teknologi MARA, Malaysia and the other, from the National University of Malaysia. The first, an Associate Professor specialised in Human Resource Management while the second specialised

in

Organisational

Behaviour.

The

researcher

considered

their

specialisations as closely related to the field of Human Resource Development and therefore, they were suitable to be the panel judges. They were invited to give their comments not only about the validity of the items but also the general appearance of the questionnaire as well. The comments made by the two panel judges and the type of action taken were shown in Table 3.3 below. Table 3.3 Experts Comments Comment Expert 1 and 2: The respondent’s name, email and contact number should not be included in the demographic information, as these will cause bias.

Action Taken The researcher gained confirmation by the course co-ordinator at the training centre that this information could be obtained from the registration database. Therefore, action was taken to delete them from the questionnaire.

Expert 1: There were two items in the questionnaire that could not be answered by respondents at the conclusion of a training program.

These two items were made in future tense.

Expert 1 and 2: The length of the questionnaire (117 items).

The researcher maintained the length of the questionnaire until the questionnaire was piloted in order to gain the feedback from the respondents.

Step 7-Pilot Study According to Churchill (1979:206), data collection should never begin without an adequate pre-test of the instrument. For this reason, the questionnaire was pilot tested with 28 trainees attending a two-day workshop on Public Accounts at the training division, Accountant General’s Department of Malaysia. In the pilot test, two main factors suggested by Spector (1992:8) were examined: respondent identification of ambiguous and confusing items; and items which could not be rated using the

68

dimension chosen. Other than that, the researcher also examined the time taken by the respondents to complete all the 177 items. The department granted the permission to conduct the pilot test (see Appendix C) and respondents were informed that the study was voluntary and anyone who wished to leave was allowed to do so. All agreed to participate. The questionnaire was administered at the conclusion of the training program and collected immediately upon completion. During the pilot test, as indicated above, the respondents were concerned about the length of the questionnaire (117 items). They had been told that all items were in short sentences and should not take a long time to complete. They were also given the chance to ask any questions for clarity if they found this necessary. All the respondents said that the questionnaire was understandable and they took between 30 to 40 minutes to complete the questionnaire. The data obtained from the pilot study was then used to examine the internal consistency of the items for each construct. This is described in the next step. Step 8-Item Analysis In this step, item analysis was conducted to find those items that formed an internally consistent scale and to eliminate those items that did not (Spector 1992:29). For this reason, the researcher adopted several recommendations made by experts while conducting this step as follows: •

The item-to-total correlations exceed 0.50 and the inter-item correlations exceed 0.30 (Hair et al. 1998:118).



Reliability coefficient alpha for a new scale should be at least 0.70 (Nunnally 1978:245) or it may decrease to 0.60 in an exploratory research (Hair et al. 1998:118).



Item means close to the centre of the range of possible scores is desirable (De Vellis 2003:94).



A scale item that has relatively high variance is preferred (De Vellis 2003:93).

69

The result of the item analysis is presented in Table 3.4. Table 3.4 Results of the Item Analysis Construct

Total Items

Number of Items Retained 5 4

Cronbach’s Alpha

7 7

Number of Items Dropped 2 3

Learner readiness Performance-self efficacy Motivation to transfer Transfer effortperformance expectations Performanceoutcome expectations Feedback Peer Support Supervisor Support Openness to Change Personal Outcomes-Positive Personal Outcomes-Negative Supervisor Sanctions Personal capacity for Transfer Opportunity to Use Content Validity Transfer Design Sharing Behaviour Intention to Share Attitude toward Knowledge Sharing Subjective Norm toward Knowledge Sharing Perceived Behavioural Control toward Knowledge Sharing Total

6

2

4

0.78

6

1

5

0.68

5

1

4

0.83

5 5 5 6

1 1 1 2

4 4 4 4

0.68 0.86 0.69 0.85

6

2

4

0.78

5

1

4

0.76

5

1

4

0.90

5

2

3

0.61

5 7 6 6 5 5

1 2 2 1 1 0

4 5 4 5 4 5

0.80 0.75 0.92 0.85 0.84 0.63

5

1

4

0.85

5

2

3

0.86

117

30

87

0.66 0.86

70

As expected, several items had to be dropped due to low reliability. Nevertheless, all scales had an adequate number of items (at least three items) to achieve content adequacy (Cook et al. 1989). Two scales (personal capacity for transfer and perceived behavioural control toward knowledge sharing) had three items respectively while in other scales, items ranged from four to five items per scale. Although the Cronbach’s Alpha reliability was based on a small sample of respondents (n = 28), it still served as an indicator that the scales were consistent in measuring the intended constructs. Finally, 87 items were retained and used in the final questionnaire for data collection. The researcher maintained the wordings of all the retained items as the feedback received from the respondents during the pilot study indicated that they were understandable. Step 9-Finalising the Questionnaire In this step, the researcher repeated Step 4 (determine item sequence) and Step 5 (determine layout and appearance) as described earlier. Appendix A provides the final questionnaire used in data collection. The next section describes the sample and the data collection procedures taken in this thesis.

3.3.2 Sample and Data Collection As outlined in Chapter 1, the target population of this study were government employees attending training at the National Institute of Public Administration, which is a central training organisation for government employees in Malaysia. When this study was proposed, the researcher was located in Melbourne and accessibility to the sample was limited. Therefore, purposive and accidental sampling techniques were used because these techniques were considered as more achievable. Accidental sampling involves using available cases for a study while purposive sampling refers to sample elements judged to be typical or representative, are chosen from the population (Ary, Jacobs & Razavieh 2002:169).

71

Once a letter of approval was received from the training centre (see Appendix C) and ethics clearance for data collection was granted by Victoria University on the 21st July 2005 (see Appendix D), the survey began from August 2005 until September 2005. The questionnaire in Bahasa Malaysia version was used in the data collection process. Prior to data collection, the researcher travelled to Malaysia and arranged several meetings with training officers at the training centre as well as with the training co-ordinators to identify the training programs to be held in the two month period of data collection. In these meetings, the researcher also discussed with them the best way to administer the questionnaire without disrupting the learning process. As a result of these meetings it was agreed that: •

The researcher would be responsible for preparation of the questionnaire for each training program. The questionnaire would be given to the training co-ordinator at least a day before training was to start.



The researcher would be responsible to ensure that each questionnaire set contained at least 25 copies as no class was to have more than 25 trainees. However, in certain circumstances, the number may reach to sixty depending on the type of training.



The researcher would also be responsible to ensure that each questionnaire set should be put into an envelope with the training name, day and date of training clearly identified on the cover.



The questionnaire would be administered at the conclusion of the training program and would be collected by the training co-ordinator immediately upon completion. The researcher would be responsible for collection of returned questionnaires the next day.



Trainees would be allowed to take back the questionnaire if they did not have the time to complete them in the class. If this happened, trainees would be asked to send back the questionnaire within two weeks to the training co-ordinator.



The training officers and the training co-ordinators would take no responsibility should there be any questionnaire missing or for incomplete returned questionnaires. 72

Six training providers at the training centre agreed to participate in this study. Unfortunately, many training programs were cancelled due to low participation and accommodation problems. Despite this, 19 types of training programs were involved and were categorised into three types: general training (for example, quality report writing); management training (for example, human resource management) and computer training (for example, visual basic). A total of 437 questionnaires were distributed. Of these, 358 were returned, representing an 82 percent response rate. After checking all the returned questionnaires, only 291 were considered usable (complete) while 67 questionnaires were incomplete on the basis that they contained more than one page unanswered and therefore had to be excluded from analysis (see Appendix E for the number of distributed and returned questionnaires). Thus, the effective return rate for the questionnaires was 66.5 per cent. Despite being quite a bit lower than the initial 82 percent return rate, it is still a high rate of return for questionnaire administration which can in part be attributed to the effectiveness of the agreed list of responsibilities between researcher and training providers.

73

3.3.3 Analysis Strategy This section describes the analysis strategy undertaken for data screening, checking for outliers, checking the multivariate assumptions, examining construct validity and reliability and hypothesis testing. Data Screening Once the data were entered, it was screened to ensure that no errors in data entry had occurred as clearly, these can distort the statistical analyses. This was done by detecting any ‘out of range values’ using the ‘Descriptive’ and ‘Frequencies’ commands using SPSS version 15 statistical software. Further, all negative-worded items were reversed scored so that higher scores indicate higher levels of agreements (Coakes & Steed 2003; Pallant 2005)(see Appendix F for the reversed scored items). Checking for Outliers Outliers refer to a substantial difference between the actual value for the dependent or independent variable and the predicted value (Hair et al. 1998). It may occur due to error in data entry. Therefore, the researcher checked the casewise diagnostics to detect cases that have standardised residual values above 3.0 or below –3.0 (Pallant 2005). If any case is found, Cook’s Distance value in the residuals statistic table will be checked in order to know whether the strange cases are having any undue influence on the regression results. According to Pallant (2005), any value larger than 1.0 is a potential problem and the cases should be considered for removal. In this thesis, the casewise diagnostics has identified one case (case number 58) with a residual value –3.596, which was below –3.0 (Pallant 2005). Further investigation found that, the respondent for case number 58 recorded a total motivation to transfer training score of 2.8, but the predicted value was 3.99, indicating that the respondent was less motivated than predicted. An inspection on the value of Cook’s Distance indicated that the value was 0.27, which was less than 1.0. Thus, it was not considered as a major problem (Pallant 2005) and the case number 58 was retained.

74

Checking the Multivariate Assumptions: Multicollinearity, Normality, Linearity, Homoscedasticity and Independence of Residuals. According to Hair et al. (1998), researchers should check the assumptions underlying multivariate analysis before any statistical analysis is undertaken to ensure that they are

met.

These

assumptions

are

multicollinearity,

normality,

linearity,

homoscedasticity and independence of residuals. The checking of these assumptions is described below. First, the researcher checked for the impact of multicollinearity which refers to the relationship among the independent variables (Hair et al. 1998:156). The presence of multicollinearity is not desirable because as it increases, the predictive power of the independent variables decreases (Tabachnick & Fidell 2001). The following assessments were made to determine whether multicollinearity existed in this study. First, the correlation matrix for all variables was checked. A correlation above 0.90 is the first indicator of multicollinearity (Hair et al. 1998). In this thesis, the correlation matrix indicated that all the correlations were below 0.90 and thus, multicollinearity was not a problem (see Table 3.5 for the correlation matrix).

75

Table 3.5 Correlation Matrix Variables Y1:Motivation to transfer

Mean

SD

Y1

X1

X2

X3

X4

X5

X6

X7

X8

X9

X10

X11

X12

X13

X14

X15

X16

X17

X18

X19

X20

4.334 0.486 0.81

X1:Opportunity to use

3.444 0.734 0.29** 0.83

X2:Supervisor sanctions

2.233 0.870 -0.25** -0.05 0.87

X3:Transfer design

4.172 0.489 0.33** 0.19** -0.19** 0.86

X4:Content validity

4.029 0.567 0.46** 0.33** -0.36** 0.32** 0.78

X5:Personal outcomes-negative

3.200 0.827 0.25** 0.28** 0.10

X6:Personal capacity for transfer

4.047 0.521 0.46** 0.23** -0.15* 0.30** 0.38** 0.26** 0.80

0.17** 0.13* 0.79

X7:Personal outcomes-positive

4.013 0.529 0.65** 0.30** -0.18** 0.36** 0.35** 0.32** 0.41** 0.72

X8:Supervisor support

3.791 0.637 0.33** 0.53** 0.24** 0.26** 0.30** 0.25** 0.31** 0.41** 0.86

X9:Peer support

3.589 0.638 0.31** 0.51** -0.11

X10:Learner readiness

4.422 0.528 0.66** 0.22** -0.25** 0.29** 0.34** 0.14* 0.36** 0.49** 0.28** 0.27** 0.73

X11:Transfer effort-performance expectations X12:Openness to change

4.054 0.540 0.45** 0.46** -0.18** 0.24** 0.59** 0.31** 0.35** 0.45** 0.48** 0.48** 0.40** 0.85 2.745 0.867 -0.07

-0.10 0.41** -0.08 -0.09 0.15** -0.10 -0.15** -0.17** -0.18** -0.11 -0.08 0.93

X13:Performance-outcomes expectations X14:Performance-self efficacy

3.385 0.659 0.14*

0.39** 0.04

X15:Feedback

3.808 0.603 0.22** 0.34** -0.09

X16:Sharing behaviour

3.944 0.583 0.31** 0.31** -0.12* 0.35** 0.24** 0.20** 0.36** 0.33** 0.42** 0.46** 0.24** 0.35** -0.10

0.31** 0.31** 0.34** 0.45** 0.39** 0.62** 0.88

0.12* 0.05

0.25** 0.23** 0.33** 0.35** 0.34** 0.06

0.33** 0.01

4.168 0.479 0.42** 0.15* -0.14* 0.30* 0.32** 0.12* 0.56** 0.37** 0.27** 0.31** 0.36** 0.40** -0.05

0.79 0.29** 0.83

0.29** 0.18** 0.24** 0.25** 0.44** 0.49** 0.41** 0.18** 0.35** -0.16** 0.39** 0.40** 0.81 0.30** 0.48** 0.47** 0.83

X17:Attitude toward knowledge 4.365 0.498 0.37** 0.16** -0.24** 0.31** 0.39** 0.13* 0.43** 0.29** 0.23** 0.26** 0.47** 0.44** -0.12* 0.11 0.60** 0.30** 0.41** 0.91 sharing X18:Perceived behavioural control 4.095 0.547 0.29** 0.18** -0.10 0.32** 0.19** 0.22** 0.46** 0.23** 0.23** 0.28** 0.29** 0.32** -0.02 0.27** 0.52** 0.26** 0.52** 0.51** 0.77 X19:Subjective norms

3.979 0.599 0.26** 0.23** -0.25** 0.33** 0.25** 0.20** 0.33** 0.22** 0.39** 0.35** 0.33** 0.36** -0.06

0.25** 0.47** 0.35** 0.56** 0.57** 0.58** 0.79

X20:Intention to share

4.165 0.459 0.41** 0.22** -0.21** 0.38** 0.36** 0.23** 0.48** 0.35** 0.31** 0.36** 0.39** 0.49** -0.05

0.21** 0.60** 0.37** 0.68** 0.62** 0.63** 0.58** 0.86

Note: * correlation is significant at the 0.05 level; ** correlation is significant at the 0.01 level. Coefficient alphas (α) are shown in bold italic and are located along the diagonal.

76

Next, the assumption of normality was checked. The assumption of normality is that errors of prediction are normally distributed about the predicted dependent variable score (Tabachnick & Fidell 2001:119). This assumption was checked through a normal probability plot, which compared the standardised residuals with the normal distribution. The normal distribution forms a straight diagonal line and the plotted residual values are compared with the diagonal. In this thesis, the plotted residual value lay in a reasonably straight diagonal line from bottom left to top right, indicating that the assumption of normality was met (Hair et al. 1998; Tabachnick & Fidell 2001; Pallant 2005) (see Appendix H). Finally,

checking

the

multivariate

assumptions

was

made

for

linearity,

homoscedasticity and independence of residuals simultaneously. The linearity of the relationship between dependent and independent variables represents the degree to which the change in the dependent variable is associated with the independent variable (Hair et al. 1998:173). Non-linear effects would result an underestimation of the actual strength of the relationship because correlations represent only the linear association between variables. On the other hand, the assumption of homoscedasticity refers to the assumption that dependent variable exhibit equal levels of variance across the range of predictor variable (Tabachnick

& Fidell

2001:79).

Homoscedasticity is desirable because the variance of the dependent variable being explained in the dependence relationship should not be concentrated in only a limited range of the independent values (Hair et al.1998:175). Further, independence of residual refers to the assumption that the residuals have a linear relationship with the predicted dependent variable scores, and that the variance of the residuals is the same for all predicted scores (Hair et al. 1998). These assumptions were checked by examining a scatterplot of the standardised residuals. The scatterplot indicated that the scores concentrated in the centre (along the 0 point), indicating that it met the assumptions for linearity, homoscedasticity and independence of residuals (Hair et al. 1998; Tabachnick & Fidell 2001; Pallant 2005) (see Appendix I).

77

Construct Validity Validity is defined as the extent to which any measuring instrument measures what it is intended to measure (Kerlinger 1986:417). In this thesis, construct validity was examined through both content and construct validity. As content validity has been described in the questionnaire design section (see section 3.3.1, step 6), the discussion in this section is limited to the statistical analysis undertaken to examine construct validity. The survey instrument was factor analysed using both exploratory factor analysis (principal component analysis) and confirmatory factor analyses using structural equation modelling to filter the best items that can represent the constructs under study. Exploratory factor analysis was used in this thesis to confirm the dimensions of the concepts that have been operationally defined as well as to indicate which of the items were most appropriate for each dimension (Hair et al. 1998; Hurley, Scandura, Schriesheim, Brannick, Vandenberg & Williams 1997; Spector 1992). Confirmatory factor analysis was then used because it provides the measurement error and a measure of model fit (Hair et al. 1998; Tabachnick & Fidell 2001). In transfer of training research, the use of exploratory and confirmatory factor analysis to examine construct validity was found in two studies (Naquin & Holton 2003; Tracey et al. 1995). However, according to Chin (1998), it should be regarded as exploratory when items were dropped due to poor wordings in order to increase the validity of the scale. In this thesis, three items were dropped due to poor wording. By dropping these items the validity of the constructs increased (see Chapter 4, section 4.2.4; 4.2.17; 4.2.19). In exploratory factor analysis, the technique used to retain factors was the latent root criterion. This technique retained only factors having eigenvalues greater than 1 while factors having eigenvalues less than 1 were considered insignificant and disregarded (Hair et al. 1998). Further, varimax rotation was applied to increase the interpretability of factor rotation (Hair et al. 1998). Simple structure is deemed to be attained if each of the original items load on one, and only one factor (De Vellis 2003). Statistical significance of item loadings was assessed using the guidelines

78

recommended by a number of researchers (Ford, MacCallum & Tait 1986; Hair et al. 1998). For example, Ford et al. (1986) suggested that only items with loadings greater than ± 0.40 are considered significant and used in defining factors. Hair et al. (1998:112) provided clearer guidelines for identifying significant item loading based on sample size. As the sample size used in this thesis was n=291, the cut-off point chosen for item loading was 0.35 and any items below this cut-off were not displayed in the results (Hair et al. 1998). Then, the items retained in the exploratory factor analysis (principal component analysis) were submitted to a confirmatory factor analysis (structural equation modelling) using AMOS version 7 statistical software. In a confirmatory factor analysis, the measurement model for each construct was created. A measurement model specifies the relations of the observed measures to their posited underlying constructs (Anderson & Gerbing 1988). The Maximum Likelihood (ML) method was chosen to estimate the difference between the observed and estimated covariance matrices because it is the most common procedure with a sample size above 150 and efficient when the assumption of multivariate normality is met (Anderson & Gerbing 1988; Hair et al. 1998; Tabachnick & Fidell 2001). As described above, the sample size utilised here was n=291 and the assumption of normality was not violated (see Appendix H). Thus, the ML method was considered justified. The measurement models were evaluated by examining the factor loading/regresssion weight of each item for statistical significance. The factor loading should be at least 0.50 and above for adequate individual item reliability (Bagozzi & Yi 1988). Thus, in this study the consideration to drop items was made if the factor loading for each item was below the recommended level 0.50. Further, the squared multiple correlation for each item shows the amount of variance captured by each item. The closer the value to 1.0, the better the item acts as an indicator of the construct (Diamantopoulos 1994). Then, the construct’s reliability and variance extracted were calculated using the formula provided by Hair et al. (1998) as stated below:

79

Construct reliability (Sum of standardised loadings)2 =

____________________________________________

(Sum of standardised loadings)2 + Sum of indicator measurement error * Variance extracted Sum of squared standardised loadings =

___________________________________________________

Sum of squared standardised loadings + Sum of indicator measurement error * Sum of indicator measurement error = 1 – (standardised loading)2

(Source: Hair et al. 1998:624) In this study, the cut-off value for construct reliability was 0.70 for the items to be considered as sufficient in representing the constructs (Hair et al. 1998). On the other hand, construct validity was obtained when the amount of variance extracted by the construct in relation to the amount of variance due to measurement error exceeds 0.50 (Fornell & Larcker 1981). A mixture of fit indices was employed to assess the overall fit of the measurement models. The χ2 statistic was used to measure the overall fit of the measurement models. In this study, the researcher looked for non-significant differences (p>0.05 or p>0.01) because the test was between the actual and predicted matrices (Hair et al. 1998). However, because the χ2 statistic is sensitive to both small and large sample sizes (Kline 2005; Hair et al. 1998; Joreskog & Sorbom 1993), this study complements the χ2 statistic with other goodness-of-fit measures such as the Goodness of Fit Index (GFI), the Adjusted Goodness of Fit Index (AGFI), the Standardised Root Mean Square Residual (RMSR), the Tucker Lewis Index (TLI),

80

the Comparative Fit Index (CFI), the Normed Fit Index (NFI) and the Root Mean Square Error of Approximation (RMSEA) (Hair et al. 1998). The recommended value for GFI, AGFI, TLI, CFI and NFI is 0.90 or greater where value less than 0.90 considered as poor fit (Hair et al. 1998). The RMSR should have value less than 0.10 where value equal to or greater than 0.10 would indicate poor fit (Kline 2005). The recommended value for RMSEA should be no more than 0.08 for reasonable error of approximation (Kline 2005; Hair et al. 1998). When the measurement model did not show a good fit, the modification indices provided by AMOS were examined. The modification indices show the predicted decreased in χ2 if items representing a construct are allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993). Items were allowed to correlate when inspection found that they were redundant due to poor wording. Construct Reliability Reliability refers to the precision of measurement (Roscoe 1975:130). Reliability is synonymous with other terms such as dependability, stability, consistency, predictability and accuracy (Kerlinger 1986:404). According to Nunnally (1978:229). Investigations of reliability should be made when new scales are developed. There are two key aspects of reliability: consistency of the items within a scale and stability of the scale over time (Hinkin 1995). Consistency of items (internal consistency) within a scale refers to the homogeneity of the items in the scale that tap the construct while stability refers to the ability of a scale to remain the same over time or yield the same results on repeated trials (Carmines & Zeller 1979:11). In this thesis, the stability of the scale was not examined because the researcher encountered difficulty in obtaining the same group of people after a period of time. Therefore, the thesis only examined the consistency of the scale. Cronbach’s coefficient alpha was used to examine the consistency of the entire scale (Nunnally 1978; Carmines & Zeller 1979; DeVellis 2003). The guideline provided by DeVellis (2003:95) was used to check the consistency of the entire scale:

81

below 0.60 (unacceptable); between 0.60 and 0.65 (undesirable); between 0.65 and 0.70 (minimally acceptable); between 0.70 and 0.80 (respectable); between 0.80 and 0.90 (very good); and for values above 0.90, one should consider shortening the scale. In this study, the desired cut-off for Cronbach’s alpha was 0.70 because this value is the accepted level of internal consistency reliability (Carmines & Zeller 1979; De Vellis 2003). Further, internal consistency was also assessed by looking at the itemto-total correlation (the correlation of the item to the summated scale score), and the inter-item correlation (the correlation among items). Research has suggested that the rule of thumb for item-to-total correlation is above 0.50 and above 0.30 for inter-item correlations (Hair et al. 1998). Thus, in this study the desired cut-off for item-to-total correlation and inter-item correlation was 0.50 and 0.30 respectively. Therefore, any items below the cut-off values were dropped from analysis. The three items that were dropped as described earlier had item-to-total correlations and inter-item correlations below the desired cut-off (see Chapter 4, section 4.2.4; 4.2.17; 4.2.19). Hypothesis Testing This section describes the statistical techniques chosen for this study to test Hypotheses 1 to 10 in order to answer the six research questions in this thesis. This will be described below. Testing Hypothesis 1 (H1): H1: These transfer of training variables: motivation to transfer; secondary influences (performance-self efficacy, learner readiness); expected utility (transfer effortperformance expectations, performance-outcomes expectations); transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomespositive, personal outcomes-negative, supervisor sanctions); ability (personal

82

capacity for transfer, opportunity to use), enabling (content validity, transfer design) and TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across different training types (general training, management/leadership training, computer training). Multivariate analysis of variance (MANOVA) was used to test H1 because this hypothesis involved one categorical independent variables (training types: general training, management/leadership training, computer training) and more than one dependent variables (all the variables in the research framework are dependent variables) (Coakes & Steed 2003; Pallant 2005). In MANOVA analysis, researcher should check the equivalence of covariance matrices across groups (an example here is training types). According to Hair et al. (1998), a violation of this assumption has minimal impact if the groups are approximately of equal size (for example, if the largest group size divided by the smallest group size is less than 1.5). If it is more than 1.5, than researcher should check Box’s Test (output generated from MANOVA) for equality of covariance matrices and the significant value should be larger than 0.001 (non-significant difference) to indicate the equality of covariance matrices across groups (Coakes & Steed 2003; Pallant 2005). However, using the Box’s Test is not recommended by Harris (1985) because the test is overly powerful and it is likely to be established for large sample sizes. In transfer of training research, one study was found to have such problem with Box’s Test (Yamnill 2001). In Yamnill’s (2001) study, Box’s test was found to be significant, however, because the Box’s test was not a robust test (Harris 1985), the researcher ran the MANOVA analysis and the results were reported. It was decided here to report the Box’s Test results and run the MANOVA analysis whether or not the Box’s Test is statistically significant. The researcher also checked the Levene’s test of equality of error variance to ensure that the homogeneity of variance has not

83

been violated. The test should be non-significant to meet the assumption (Coakes & Steed 2003; Pallant 2005). In order to know whether the variables in H1 are differ in terms of their mean scores across the three different training types, the Wilks’ Lambda value and its associated significance level were checked. Generally, if the significance level is less than 0.05, then the researcher can conclude that there is a difference in terms of the variables’ means in H1 across training types. The smaller the Wilks’ Lamda value, the bigger is the difference (Coakes & Steed 2003; Pallant 2005). Then, in order to investigate further whether all the variables differ in terms of their means or just some, the tests of between-subjects effects output box was checked. In this study, the significant value utilised was p < 0.002 (0.05 divided by 21 variables) (Coakes & Steed 2003; Pallant 2005). Thus, the researcher considered the results significant only if the significant value less than 0.002. This significant value was also used for variables that were significant but at the same time violated the assumption of homogeneity of variance using the Levene’s test (p < 0.01). The assumption is violated when the significant value for Levene’s Test is less than 0.01 (Coakes & Steed 2003; Pallant 2005). According to Coakes & Steed (2003), if the assumption is violated, researchers must interpret finding at a more conservative alpha level. Thus, the significant value less than 0.002 was used. Finally, the researcher checked the estimated marginal means output box in order to know which variables had higher and lower scores (Coakes & Steed 2003; Pallant 2005). It was decided here that the mean scores equal to or above 4.0 would be considered a strong level; between 3.5 and 4.0, moderate, and below 3.50, poor. It should be noted here that Likert point scale ranging from 1Strongly Disagree to 5-Strongly Agree was utilised. Testing Hypothesis 2 (H2): H2: These transfer of training variables: motivation to transfer; secondary influences (performance-self efficacy, learner readiness); expected utility (transfer effortperformance expectations and performance-outcomes expectations); transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-

84

positive, personal outcomes-negative, supervisor sanctions); ability (personal capacity for transfer and opportunity to use), enabling (content validity and transfer design) and TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing and perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across trainees’ demographics (gender, age, level of education, work experience, position of employment). Multivariate analysis of variance (MANOVA) was used to test H2 because this hypothesis involved six categorical independent variables (gender, age, level of education, work experience, position of employment) and more than one dependent variables (all the variables in the research framework are dependent variables) (Coakes & Steed 2003; Pallant 2005). However, the six categorical independent variables were analysed separately one by one using one-way MANOVA (one categorical independent variable and all the variables in the research framework will be the dependent variables). The procedures in this analysis were identical to those described for testing H1. Testing Hypothesis 3 (H3) to Hypothesis 7 (H7): H3: Secondary influences variables (performance-self efficacy, learner readiness) will explain a significant proportion of variance in motivation to transfer. H4:

Expected

utility

variables

(transfer

effort-performance

expectations,

performance-outcomes expectations) will explain a significant proportion of variance in motivation to transfer. H5: Transfer climate variables (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions) will explain a significant proportion of variance in motivation to transfer.

85

H6: Enabling variables (content validity, transfer design) will explain a significant proportion of variance in motivation to transfer. H7: Ability variables (personal capacity for transfer, opportunity to use) will explain a significant proportion of variance in motivation to transfer. Multiple regression analysis was used to test H3, H4, H5, H6 and H7 because these hypotheses involved one dependent variable and more than one independent variable (Coakes & Steed 2003; Pallant 2005). For H3, the independent variables were performance self-efficacy and learner readiness while motivation to transfer was the dependent variable. For H4, the independent variables were transfer effortperformance expectations and performance-outcomes expectations while motivation to transfer was the dependent variable. For H5, the independent variables were feedback, peer support, supervisor support, openness to change, personal outcomespositive, personal outcomes-negative and supervisor sanctions while motivation to transfer was the dependent variable. For H6, the independent variables were content validity and transfer design while motivation to transfer was the dependent variable. For H7, the independent variables were personal capacity for transfer and opportunity to use while motivation to transfer was the dependent variable. As described earlier, in multiple regression analysis, multicollinearity is not desirable because as it increases, the predictive power of the independent variables decreases (Hair et al. 1998). Therefore, the researcher checked the tolerance value and its inverse, the variation of inflation factor (VIF) (output generated from multiple regression analysis). Tolerance represents the amount of variability of the selected independent variable not explained by the other independent variable. Thus, very small tolerance values (and thus, large VIF values) denote high collinearity (Coakes & Steed 2003; Pallant 2005). In this study, the cut-off point for determining the presence of multicollinearity was less than 0.10 for the tolerance value and above 10 for the VIF (Hair et al. 1998).

86

Testing Hypothesis 8 (H8) H8: Intention to share will be significantly correlated to sharing behaviour and sharing behaviour will be significantly correlated to motivation to transfer. H8 was tested using the Pearson correlation because this hypotheses involved two variables to examine the relationship between intention to share and sharing behaviour and two variables to examine the relationship between sharing behaviour and motivation to transfer (Coakes & Steed 2003; Pallant 2005). When examining the strength of the relationship between intention to share and sharing behaviour and between sharing behaviour and motivation to transfer, the researcher used the guidelines provided by Cohen (1988) as stated below. r = 0.10 to 0.29 small r = 0.30 to 0.49 medium r = 0.50 to 1.00 larger Testing Hypothesis 9 (H9) H9: Attitude, subjective norms and perceived behavioural control toward knowledge sharing will explain a significant proportion of the variance in intention to share. H9 was tested using multiple regression analysis. The procedures in this analysis were identical to those described for testing H3, H4, H5, H6 and H7. Testing Hypothesis 10 (H10) H10: Sharing behaviour will have a direct and indirect relationship (via the significant predictors identified in research question three) with motivation to transfer. H10 was tested using structural equation modelling (SEM) because of its advantage to examine the direct and indirect relationships simultaneously (via the significant predictors identified in research question three) with motivation to transfer, which

87

cannot be done using multiple regression analysis or correlational analysis (Hair et al. 1998). When modelling with SEM, the sample size required, according to Tabachnick and Fidell (2001:660) is 10 subjects per estimated parameter. Due to the sample size used in this study (n = 291), the researcher was unable to include all the significant predictors identified in research question three into the structural model. Therefore, the significant predictors that had the strongest beta value from each category of variables were selected into the structural model (see Chapter 5; section 5.3.6). When developing the structural model, the regression coefficient (λ) and the measurement error variances (θ) for each variable in the structural model were calculated as suggested by Politis (2001; 2002; 2003) and Joreskog and Sorbom (1989). The regression coefficient reflects the regression of each composite variable on its latent variable and the measurement error variance (θ) associated with each composite variable (Politis 2001; 2002; 2003). In this study, where the matrix to be analysed was a matrix of covariances (the correlation and its variance) amongst the composite variables, then λ and θ were calculated using the equations stated below (see Appendix N for the calculation of λ and θ for each variable included in the structural model). λ = σ√α θ = σ2(1−α) where: λ = regression coefficients θ = measurement error variances α = reliability coefficient σ = standard deviation of composite measure; and σ2 = error variance (Source: Politis 2001:359) In turn, these values have been used as fixed parameters in the structural model as shown in the simplified structural model of Figure 3.2. 88

Figure 3.2 Simplified Structural Model

θ

X

λ

Sharing behaviour

(ξ)

γ

Content validity

λ

Y

θ

(η)

Where: X and Y = composite latent variables derived from measurement model λ = regression coefficients computed by equation 3 θ = measurement error variances computed by equation 4 γ = the regression coefficient of the regression of η on ξ (Source: Politis 2001) The fit indices (χ2; GFI; AGFI; RMSR; TLI; CFI; NFI; RMSEA) and the desired cutoff used to assess the overall fit of the structural model were similar to those described in section 3.3.3 when evaluating measurement models for construct validity. When the structural model does not fit, it has become common practice to modify the model by deleting parameters that are not significant and adding the parameters that improve the fit (Hair et al. 1998). In this study adding or deleting parameters from the structural model were made based on theoretical justification or common sense and not based solely on the modification indices (Hair et al. 1998; Joreskog & Sorbom 1993; Arbuckle & Wothe 1999) (see Chapter Five; section 5.3.6). When examining the strength of the relationships in the structural model, the researcher used the guidelines provided by Kline (2005) which assists in the interpretations of path coefficients (λ), with small, medium or large effects. Standardised path coefficients with values less than 0.10 indicate a small effect. Values around 0.30 indicate a medium effect and values above 0.50 indicate a large effect. Further, the squared multiple correlations indicate the variance explained in the endogenous constructs (outcome constructs) accounted by its predictors

89

(Arbuckle & Wothe 1999; Joreskog & Sorbom 1993). The standardised path coefficients, the squared multiple correlations and the interpretation guidelines offered by Kline (2005) were used when examining the results.

3.4 Summary This chapter described the development of the conceptual framework and the methodology chosen to test the relationships hypothesised in the conceptual framework. Based on the conceptual framework, this thesis argued that secondary influence variables (learner readiness, performance self-efficacy), expected utility variables

(transfer

effort-performance

expectations,

performance-outcomes

expectations), transfer climate variables (feedback, peer support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions), enabling variables (content validity, transfer design) and ability variables (personal capacity for transfer, opportunity to use) would have a direct influence on motivation to transfer. Further, this thesis contributes to understanding of motivation to transfer learning by adding the variables pertaining to sharing behaviour which was hypothesised as being linked to motivation to transfer. The variables depicted in the conceptual framework were measured using a multiple items questionnaire. This chapter described the framework for questionnaire design as well as the steps taken to produce item generation, content validity, pilot testing and finalising the questionnaire. The sample chosen comprised 291 trainees from the Malaysian public sector and the procedures undertaken for data collection were also described. The chapter then moved to a discussion of the analysis strategy for data screening, checking for outliers and the multivariate assumptions, examining construct validity using exploratory factor analysis (principal component analysis) and examining construct reliability using Cronbach’s alpha. Finally, the statistical techniques used to test the hypotheses formulated in this thesis were described. These included multivariate analysis of variance (MANOVA), multiple regression analysis, Pearson correlation and structural equation modelling.

90

The next chapter continues the methodology for this study with a focus on construct validity using exploratory factor analysis (principal component analysis) and confirmatory factor analysis (structural equation modelling). Construct reliability using Cronbach’s alpha is also described in the next chapter.

91

CHAPTER 4 CONSTRUCT VALIDITY AND RELIABILITY

4.1 Introduction In Chapter 3, the development of the conceptual framework was described along with the methodology chosen to examine the 10 hypotheses. The chapter also detailed the procedures undertaken to examine construct validity and reliability using exploratory factor analysis (principal component analysis), confirmatory factor analysis (structural equation modelling) and Cronbach’s alpha. This chapter continues the discussion on methodology through a presentation of the results of construct assessment using these techniques. The chapter confirms that all constructs measure what they intended to measure and display good psychometric properties (validity and reliability) to investigate the relationships hypothesised in this thesis.

4.2 Construct Assessment Construct assessment was conducted and items for each construct were analysed individually using exploratory factor analysis (principal component analysis) and Cronbach’s alpha for internal consistency reliability. A summary of the construct assessment is provided below (a more detailed discussion was provided in Chapter 3): In the exploratory factor analysis: -

Statistical Package for Social Sciences (SPSS) version 15 was utilised.

-

All constructs were factor analysed with principal component analysis (varimax rotation) (Hair et al. 1998).

92

-

Only factors having eigenvalue above 1 were considered significant and retained (Hair et al. 1998).

-

The cut-off point for item loading was 0.35 and any items below the desired cutoff were not displayed in the results (Hair et al. 1998).

-

The consistency among the items was checked using Cronbach’s alpha (desired cut-off was 0.70) (Carmines & Zeller 1979; De Vellis 2003). In addition, internal consistency among the items was also checked by looking at the item-to-total correlation (desired cut-off 0.50) and inter-item correlation (desired cut-off 0.30) (Hair et al. 1998). Any items below the desired cut-off for Cronbach’s alpha, item-to-total correlation and inter-item correlation respectively were dropped.

In the confirmatory factor analysis: -

The Analysis of Moment Structure (AMOS) version 7 was utilised.

-

The measurement models for each construct were constructed.

-

The measurement models were evaluated by examining the factor loading for each item. The desired cut-off for factor loading was taken to be at least 0.50 for adequate individual item reliability (Bagozzi & Yi 1988).

-

The χ2 statistic was used to measure the overall fit of the measurement models. In this thesis, the researcher looked for non-significant differences (p>0.05 or p>0.01) because the test was between the actual and predicted matrices (Hair et al. 1998).

-

Other fit indices utilised were Goodness of Fit Index (GFI), the Adjusted Goodness of Fit Index (AGFI), the Standardised Root Mean Square Residual (RMSR), the Tucker Lewis Index (TLI), the Comparative Fit Index (CFI), the Normed Fit Index (NFI) and the Root Mean Square Error of Approximation (RMSEA) (Hair et al. 1998). The recommended value for GFI, AGFI, TLI, CFI and NFI is 0.90 or greater where value less than 0.90 considered as poor fit (Hair et al. 1998). The RMSR should have value less than 0.10 where value equal to or greater than 0.10 would indicate poor fit (Kline 2005). The recommended value for RMSEA should be no more than 0.08 for reasonable error of approximation (Kline 2005; Hair et al. 1998).

93

-

When the measurement model did not show a good fit, the modification indices provided by AMOS version 7 were examined. Items were allowed to correlate when inspection found that they were redundant due to poor wording (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993).

-

Construct reliability and variance extracted were calculated using the formula provided by Hair et al. (1998) (see Chapter 3). The desired cut-off was 0.70 and 0.50 for construct reliability and variance extracted respectively (Fornell & Larcker 1981).

The next section details the outcomes of the tests conducted on each construct. In the first stage, principle component analysis was conducted followed by a reliability analysis and finally the measurement model for each construct was developed. In total there were five learner readiness items; four performance-self efficacy items; four motivation to transfer items; five transfer effort-performance expectations items; four performance-outcome expectations items; four feedback items; four peer support items; four supervisor support items; four openness to change items; four personal outcomes-positive items; four personal outcomes-negative items; four supervisor sanctions items; three personal capacity for transfer items; four opportunity to use items; five content validity items; four transfer design items; five sharing behaviour items; four intention to share items; five attitude toward knowledge sharing items; four subjective norm toward knowledge sharing items; and three perceived behavioural control toward knowledge sharing items. A detailed description of the definition of each construct was presented in Table 1.1 in Chapter 1.

94

4.2.1 The Learner Readiness Construct Principal component analysis with varimax rotation was conducted on the five learner readiness items. As expected, only one factor was extracted (eigenvalue above 1) which accounted for approximately 55.36 percent of the total variance, confirming the unidimensionality of this construct (De Vellis 1991; Hair et al. 1998). All items had factor loading above the recommended cut-off 0.35 (Hair et al. 1998) (see Table 4.1). Therefore, no items were dropped following this stage of item testing. Table 4.1 Learner Readiness Principal Component Analysis Items Q1. I know that this training is good for me. Q2. I applied for this training on my own. Q5. I am definitely interested to join this training. Q8. I am definitely ready to join this training. Q9. I knew that I would obtain something beneficial from this training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor : Learner Readiness 0.66 0.58 0.83 0.80 0.81 291 2.77 55.36 0.73

Reliability analysis was also conducted with these items. The results indicated that all items had item-total correlation above 0.50 except two items (Q1=0.47 and Q2=0.41) (see Appendix J) which were slightly below the recommended level of 0.50. Further inspection on the inter-item correlation matrix revealed that all items were above the recommended cut-off, 0.30 (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha was calculated at 0.73, also above the recommended cut-off, 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, no items were dropped for this scale. Next, the measurement model for the learner readiness construct was constructed with the five items as depicted in Figure 4.1.

95

Figure 4.1 Measurement Model: Learner Readiness Construct .29

e1

q1 .20

e2

.54

q2

.45

.66

e3

.81

q5

.58 .76

e4

q8

Learner readiness

.74 .55

e5

q9

Table 4.2 Fit Indices for Learner Readiness Construct χ2 value Degrees of freedom p value GFI AGFI

7.519 5 0.185 0.989 0.967

RMSR TLI CFI NFI RMSEA

0.011 0.988 0.994 0.983 0.042

The measurement model for the learner readiness construct produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items were also above the recommended cut-off, 0.50 indicating adequate individual item reliability (Bagozzi & Yi 1988). There was one exception: q2=0.45. However, despite the exception, this item was retained because the fit indices supported a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off, 0.90 (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08 respectively (Kline 2005) (see Table 7.23). The construct reliability was above the recommended level 0.70 (Hair et al. 1998) and the variance extracted by the construct was slightly below the recommended 0.50 cut-off due to measurement error (Fornell & Larcker 1981) (see Table 4.3). Despite the

96

slightly lower variance extraction, principal component analysis with varimax rotation revealed that the variance extracted was 55.36 percent, clearly above the 50 percent, indicating that this construct possessed adequate reliability and validity. Table 4.3 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Learner Readiness Construct Construct

Mean

Standard Deviation (δ δ) 0.528

Cronbach’s Alpha (α α) 0.731

Construct Reliability

Variance Extracted

Motivation 4.422 0.80 to transfer Note: see Appendix K for construct reliability and variance extracted workings.

0.455

4.2.2 The Performance-Self Efficacy Construct Principal component analysis with varimax rotation was conducted on the four performance-self efficacy items. Results indicated that only one factor was extracted (eigenvalue above 1), confirming the unidimensionality of this construct (De Vellis 1991; Hair et al. 1998). The percentage of variance extracted was approximately 67.16 percent. All items had factor loadings above the recommended cut-off, 0.35 (Hair et al. 1998) (see table 4.4). Therefore, no items were dropped following this stage of item testing.. Table 4.4 Performance-Self Efficacy Principal Component Analysis Items Q59. I am confidence to increase my job performance. Q60. I have the capabilities to increase my job performance. Q63. I am confident that I can improve my job performance because I am a discipline person. Q64. I am confident that I can improve my job performance because I am a hardworking person. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Performance-Self Efficacy 0.74 0.82 0.85 0.85 291 2.69 67.16 0.83

Following this, reliability analysis was conducted with these items. The results indicated that all items displayed item-total correlation and inter-item correlation above the desired cut-offs, 0.50 and 0.30 respectively (Hair et al. 1998) (see

97

Appendix J). Cronbach’s alpha was 0.83, also above the recommended cut-off, 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, no items were dropped from this scale. Next, the measurement model for performance-self efficacy construct was dimensioned with the four items as depicted in Figure 4.2 below. Figure 4.2 Measurement Model: Performance-Self Efficacy Construct .22

e1

q59 .47

.35

.53

e2

.59

q60 .82

e3

q63 .80

e4

.90

Performanceself efficacy

.89

q64

Table 4.5 Fit Indices for Performance-Self Efficacy Construct χ2 value Degrees of freedom p value GFI AGFI

0.447 1 0.504 0.999 0.992

RMSR TLI CFI NFI RMSEA

0.001 1.006 1.000 0.999 0.000

The χ2 statistic for the initial measurement model for the performance-self efficacy construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination on the modification indices showed that major improvements could be achieved if errors for Q59 and Q60 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.2). When, this was done, the model produced an acceptable fit to the data with a non-significant χ2 statistic (p>0.05), indicating that

98

the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items were also above the recommended cut-off, 0.50 (Bagozzi & Yi 1998) except for one item (q59), which was slightly below the 0.50 cut-off. However, this item was retained because the fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08 respectively (Kline 2005) (see Table 4.5). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.6). Table 4.6 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Performance-Self Efficacy Construct Construct

Mean

Standard Deviation (δ δ) 0.479

Cronbach’s Alpha (α α) 0.832

Construct Reliability

Performance 4.168 0.816 -self efficacy Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.543

4.2.3 The Motivation to Transfer Construct Principal component analysis with varimax rotation was conducted on the four motivation to transfer items. As expected, only one factor was extracted (eigenvalue above 1), again confirming the unidimensionality of this construct (De Vellis 1991; Hair et al. 1998). The percentage of variance extracted was approximately 63.67 percent. All items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.7). Therefore, no items were dropped following this stage of item testing.

99

Table 4.7 Motivation to Transfer Principal Component Analysis Items Q3. I will put into practice what I have learned from the training to the office. Q6. I will make sure that what I have learned from the training will be put into practice for job benefit. Q10. I will work as hard as possible to put into practice what I have learned for job benefit. Q12. I will do a plan to put into practice what I have learned after I get back to the offfice. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Motivation to Transfer 0.78 0.86 0.81 0.73 291 2.55 63.67 0.81

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations above 0.50 and the inter-item correlation matrix was above the 0.30 recommended values (Hair et al. 1998) (see Appendix J). Cronbach’s alpha for this scale was calculated at 0.81, also above the recommended cut-off 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, no items were dropped for this scale. Next, the measurement model for motivation to transfer construct was developed with the four items as depicted in Figure 4.3. Figure 4.3 Measurement Model: Motivation to Transfer Construct .49

e1

q3 .70

.71

e2

q6

.84 .52

e3

q10 .37

e4

.72

Motivation to transfer

.61

q12

100

Table 4.8 Fit Indices for Motivation to Transfer Construct χ2 value Degrees of freedom p value GFI AGFI

6.027 2 0.049 0.989 0.947

RMSR TLI CFI NFI RMSEA

0.009 0.968 0.989 0.984 0.083

The measurement model for motivation to transfer produced a good model fit with the data with a non-significant χ2 statistic (p>0.01), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988). Further, the fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). However, the RMSR was below the recommended level of 0.10, and the RMSEA was slightly above the recommended level of 0.08 (Kline 2005) (see Table 4.8). The construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.9). Table 4.9 Descriptive Statistics, Cronbach’s alpha, Construct Reliability and Variance Extracted for Motivation to Transfer Construct Construct

Mean

Standard Deviation (δ δ) 0.486

Cronbach’s Alpha (α α) 0.806

Construct Reliability

Motivation 4.334 0.811 to transfer Note: See Appendix K for construct reliability and variance extracted workings.

4.2.4

The

Transfer

Effort-Performance

Variance Extracted 0.522

Expectations

Construct Principal component analysis with varimax rotation was conducted on the five transfer effort-performance expectations items. As expected, only one factor was extracted (eigenvalue above 1), again confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). The percentage of variance extracted was

101

approximately 55.92 percent. This time, all items had factor loadings above the 0.35 cut-off (Hair et al. 1998) except for one item (Q53) which had a factor loading of 0.29. Further inspection found that this item was a negative worded item and it was performing poorly due to its poor wording (see Appendix F for the negative worded items). Therefore, action was taken to drop this item and the program was-rerun. Results indicated that the percentage of variance increased to approximately 68.48 percent. Further, this time, all items had factor loadings above the recommended cutoff 0.35 (Hair et al. 1998) (see Table 4.10). Table 4.10 Transfer Effort-Performance Expectations Principal Component Analysis Items Q46. I expect my work will be more efficient if I put into practice what I have learned from the training. Q47. I expect my quality of work will be better if I put into practice what I have learned from the training. Q50. I expect that my work will be more effective if I put into practice what I have learned from the training. Q51. I expect that my productivity will be increased if I put into practice what I have learned from the training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Transfer EffortPerformance Expectations 0.84 0.86 0.81 0.80 291 2.74 68.48 0.85

Then, reliability analysis was conducted with the remaining four items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs of 0.50 and 0.30 respectively (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha for this scale was calculated at 0.85, also above the recommended cut-off 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, the remaining four items were retained for this scale. Next, the measurement model for transfer effort-performance expectations construct was developed with the four items as depicted in Figure 4.4 below.

102

Figure 4.4 Measurement Model: Transfer Effort-Performance Expectations Construct .73

e1

q46 .85

.84

e2

q47

.91 .33

e3

q50 .32

.51

e4

.58

Transfer effort -performance expectations

.57

q51

Table 4.11 Fit Indices for Transfer Effort-Performance Expectations Construct χ2 value Degrees of freedom p value GFI AGFI

1.148 1 0.284 0.998 0.980

RMSR TLI CFI NFI RMSEA

0.002 0.998 1.000 0.998 0.023

The χ2 statistic for the initial measurement model for the transfer effort-performance expectations construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q50 and q51 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.4). When, this was done, the model produced an acceptable fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items also above the recommended 0.50 cut-off indicated adequate individual item reliability (Bagozzi & Yi 1998). The fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off, 0.90 (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.11). Construct reliability above the recommended level of 0.70 (Hair et al. 1998) and 103

variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.12). Table 4.12 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Transfer Effort-Performance Expectations Construct Construct

Mean

Standard Deviation (δ δ) 0.540

Cronbach’s Alpha (α α) 0.846

Construct Reliability

Variance Extracted

Transfer 4.054 0.826 effortperformance expectations Note: See Appendix K for construct reliability and variance extracted workings.

0.553

4.2.5 The Performance-Outcomes Expectations Construct Principal component analysis with varimax rotation was conducted on the four performance-outcome expectations items. As expected, only one factor was extracted (eigenvalue above 1), again confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). The percentage of variance extracted was approximately 61.46 percent. Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.13). Therefore, no items were dropped following this stage of item testing. Table 4.13 Performance-Outcomes Expectations Principal Component Analysis Items

Q48. I expect that I will receive various facilities if my job performance increased. Q49. I expect that I will be more entrusted if my job performance increased. Q52. I expect that I will be rewarded if my job performance increased. Q54. I expect that I will be promoted if my job performance increased. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: PerformanceOutcomes Expectations 0.82 0.70 0.85 0.76 291 2.46 61.46 0.79

104

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlation and inter-item correlation above the recommended cut-offs, 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha for this scale was calculated at 0.79, also above the recommended cut-off 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, no items were dropped from this scale. Then, the measurement model for performance-outcomes expectations construct was developed with the four items as depicted in Figure 4.5 below. Figure 4.5 Measurement Model: Performance-Outcomes Expectations Construct .46

e1

q48 .68

.23

.25

e2

.48

q49 .78

e3

q52 .45

e4

.88

Performanceoutcomes expectations

.67

q54

Table 4.14 Fit Indices for Performance-Outcomes Expectations Construct χ2 value Degrees of freedom p value GFI AGFI

0.860 1 0.354 0.999 0.985

RMSR TLI CFI NFI RMSEA

0.006 1.002 1.000 0.998 0.000

The χ2 statistic for the initial measurement model for the performance-outcomes expectations construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q48 and q49 were allowed to correlate (Arbuckle & Wothe

105

1999; Joreskog & Sorbom 1993) (see Figure 4.5). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended cut-off 0.50 (Bagozzi & Yi 1998) except for one item (q49). In this case, the factor loading was slightly below the 0.50 cut-off. However, this item was retained because the fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.14). Construct reliability was above the recommended level 0.70 (Hair et al. 1998). However, the variance extracted by the construct was slightly below the desired cutoff 0.50 due to measurement error (Fornell & Larcker 1981) (see Table 4.15). Despite the slightly lower variance extraction, principal component analysis with varimax rotation revealed that the variance extracted was 61.46 percent, clearly above the 50 percent, indicating that this construct possessed adequate reliability and validity. Table 4.15 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Performance-Outcomes Expectations Construct Construct

Mean

Standard Deviation (δ δ) 0.659

Cronbach’s Alpha (α α) 0.789

Construct Reliability

Performance 3.385 0.779 -outcomes expectations Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.479

4.2.6 The Feedback Construct Principal component analysis with varimax rotation was conducted on the four feedback items. As expected, only one factor was extracted (eigenvalue above 1) which accounted for approximately 64.21 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.16). Therefore, no items were dropped following this stage of item testing.

106

Table 4.16 Feedback Principal Component Analysis Items Q61. After training, I will receive feedback to improve what I have learned from the training. Q62. I will accept any constructive comment every time I put into practice what I have learned from the training. Q65. I will accept any lesson every time I try to put into practice what I have learned from the training. Q66. I will accept any good advice every time I try to put into practice what I have learned from the training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Feedback 0.69 0.75 0.87 0.88 291 2.57 64.21 0.81

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs, 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha for this scale was calculated at 0.81, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for feedback construct was constructed with the four items as depicted in Figure 4.6 below. Figure 4.6 Measurement Model: Feedback Construct .24

e1

q61 .49

.32

.24

e2

q62

.56 .76

e3

q65 .82

e4

.87

Feedback

.91

q66

107

Table 4.17 Fit Indices for Feedback Construct χ2 value Degrees of freedom p value GFI AGFI

0.937 1 0.333 0.998 0.984

RMSR TLI CFI NFI RMSEA

0.004 1.001 1.000 1.000 0.000

The χ2 statistic for the initial measurement model for feedback was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for Q61 and Q62 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.6). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended cut-off 0.50, indicating adequate individual item reliability (Bagozzi & Yi 1988). The fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.17). Construct reliability was above the recommended level 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.18). Table 4.18 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Feedback Construct Construct

Mean

Standard Cronbach’s Construct Deviation Alpha Reliability (δ δ) (α α) Feedback 3.808 0.603 0.809 0.811 Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.535

108

6.2.7 The Peer Support Construct Principal component analysis with varimax rotation was conducted on the four peer support items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 72.88 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.19). Therefore, no items were dropped following this stage of item testing. Table 4.19 Peer Support Principal Component Analysis Items Q16. My colleague will support me to put into practice what I have learned from the training. Q17. My colleague is willing to help me to put into practice what I have learned from the training. Q23. My colleague is willing to give his opinions in helping me to put into practice what I have learned from the training. Q24. My colleague will encourage me to apply what I have learned from the training in carrying out duties. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Peer Support 0.88 0.88 0.85 0.81 291 2.92 72.88 0.88

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha for this scale was calculated at 0.88, also above the recommended cut-off 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for peer support construct was constructed with the four items as depicted in Figure 4.7 below.

109

Figure 4.7 Measurement Model: Peer Support Construct .82

e1

q16 .91

.65 -.65

e2

q17

.81 .72

e3

q23 .48

e4

.85

Peer support

.70

q24

Table 4.20 Fit Indices for Peer Support Construct χ2 value Degrees of freedom p value GFI AGFI

0.391 1 0.532 0.999 0.993

RMSR TLI CFI NFI RMSEA

0.003 1.006 1.000 0.999 0.000

The χ2 statistic for the initial measurement model for the peer support construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination on the modification indices showed that major improvements could be achieved if errors for q16 and q23 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993 (see Figure 4.7). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50cut-off indicating adequate individual item reliability (Bagozzi & Yi 1988). The fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off, 0.90 (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 110

2005) (see Table 4.20). Construct reliability above the recommended level 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.21). Table 4.21 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Peer Support Construct Construct

Mean

Standard Cronbach’s Construct Deviation Alpha Reliability (δ δ) (α α) Peer support 3.589 0.638 0.875 0.891 Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.674

4.2.8 The Supervisor Support Construct Principal component analysis with varimax rotation was conducted on the four supervisor support items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 70.49 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loading above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.22). Therefore, no items were dropped following this stage of item testing. Table 4.22 Supervisor Support Principal Component Analysis Items Q25. My supervisor/manager will support me to apply what I have learned from the training in carrying out duties. Q26. My supervisor/manager will help me to apply what I have learned from the training in carrying out duties. Q29. My supervisor/manager will always provide me with encouragement to practice what I have learned from the training in carrying out duties. Q30. My supervisor/manager sets the training objective to encourage me to practice what I have learned from the training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Supervisor Support 0.90 0.89 0.84 0.73 291 2.82 70.49 0.86

111

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha for this scale was calculated at 0.86, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for the supervisor support construct was constructed with the four items as depicted in Figure 4.8 below. Figure 4.8 Measurement Model: Supervisor Support Construct .87

e1

q25 .93

.80

e2

.89

q26 .48

e3

q29 .28

.28

e4

.70

Supervisor support

.53

q30

Table 4.23 Fit Indices for Supervisor Support Construct χ2 value Degrees of freedom p value GFI AGFI

0.391 1 0.532 0.999 0.993

RMSR TLI CFI NFI RMSEA

0.003 1.006 1.000 0.999 0.000

The χ2 statistic for the initial measurement model for the supervisor support construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if

112

errors for q29 and q30 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.8). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50cut-off indicating adequate individual item reliability (Bagozzi & Yi 1988). The fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cutoff (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of: 0.10 and 0.08, respectively (Kline 2005) (see Table 4.23). Construct reliability was above the recommended 0.70 level (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.24). Table 4.24 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Supervisor Support Construct Construct

Mean

Standard Deviation (δ δ) 0.638

Cronbach’s Alpha (α α) 0.875

Construct Reliability

Supervisor 3.589 0.891 support Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.674

4.2.9 The Openness to Change Construct Principal component analysis with varimax rotation was conducted on the four openness to change items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 82.11 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.25). Therefore, no items were dropped following this stage of item testing.

113

Table 4.25 Openness to Change Principal Component Analysis Items

Factor: Openness to Change 0.89 0.95 0.94 0.84

Q55. Team is not ready to change. Q56. Team is hard to accept new ideas. Q57. Team is not ready to learn new methods. Q58. Team with more experience will not agree with any changes to be made. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

291 3.28 82.11 0.93

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). The Cronbach’s alpha for this scale was calculated at 0.93, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for openness to change construct was constructed with the four items as depicted in Figure 4.9 below. Figure 4.9 Measurement Model: Openness to Change Construct .70

e1

q55 .84

.94

e2

q56

.97 .87

e3

q57 .56

e4

.93

Openness to change

.75

q58

114

Table 4.26 Fit Indices for Openness to Change Construct χ2 value Degrees of freedom p value GFI AGFI

5.840 2 0.054 0.990 0.952

RMSR TLI CFI NFI RMSEA

0.009 0.989 0.996 0.994 0.081

The measurement model for the openness to change construct produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988). Further, the fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off 0.90 (Hair et al. 1998). The RMSR was below the recommended level 0.10, however, the RMSEA was slightly above the desired 0.08 cut-off (Kline 2005) (see Table 4.26). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.27). Table 4.27 Descriptive Statistics, Cronbach’s alpha, Construct Reliability and Variance Extracted for Openness to Change Construct Construct

Mean

Standard Deviation (δ δ) 0.867

Cronbach’s Alpha (α α) 0.925

Construct Reliability

Openness to 2.745 0.929 change Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.768

4.2.10 Personal Outcomes-Positive Construct Principal component analysis with varimax rotation was conducted on the four personal outcomes-positive items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 55.26 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.28). Therefore, no items were dropped at this stage. 115

Table 4.28 Personal Outcomes-Positive Principal Component Analysis Items Q4. I will work with more confidence if I put into practice what I have learned from the training. Q7. I will work with more organised if I put into practice what I have learned from the training. Q11. My work will be rewarded if I put into practice what I have learned. Q13. I will get a good result from the job evaluation report whenever I put into practice what I have learned from the training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Personal OutcomesPositive 0.73 0.81 0.72 0.70 291 2.21 55.26 0.72

Following this, reliability analysis was conducted. Results indicated that all items had item-total correlations above 0.50 except one item (Q4), which had and item-total correlation of 0.47 which was slightly below the 0.50 recommended value (Hair et al. 1998) (see Appendix J). Further inspection of the inter-item correlation matrix found that two correlations: Q4,Q11 and Q4,Q13 were 0.27 and 0.26 respectively, slightly below the recommended 0.30 cut-off (see Appendix J). However, because the Cronbach’s alpha for this scale was calculated at 0.72, which was above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003), all items were retained for this scale. Then, the measurement model for personal outcomes-positive construct was constructed with the four items as depicted in Figure 4.10 below. Figure 4.10 Measurement Model: Personal Outcomes-Positive Construct .44

e1

q4 .67

.89

e2

q7

.94 .17

e3

q11 .14

.44

e4

.41

Personal outcomespositive

.38

q13 116

Table 4.29 Fit Indices for Personal Outcomes-Positive Construct χ2 value Degrees of freedom p value GFI AGFI

0.119 1 0.730 1.000 0.998

RMSR TLI CFI NFI RMSEA

0.000 1.018 1.000 1.000 0.000

The χ2 statistic for the initial measurement model for personal outcomes-positive was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q11 and q13 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.10). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50 cut-off (Bagozzi & Yi 1988) except for two items (q11 and q13) (see Figure 4.10). However, these two items were retained because the fit indices: GFI, AGFI, NFI, TLI and CFI were all above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.29). Construct reliability was above the recommended 0.70 level (Hair et al. 1998), however, variance extracted by the construct was slightly below the recommended 0.50 cut-off due to measurement error (Fornell & Larcker 1981) (see Table 4.30). Despite the slightly lower variance extraction, principal component analysis with varimax rotation revealed that the variance extracted was 55.26 percent, clearly above the 50 percent, indicating that this construct possessed adequate reliability and validity.

117

Table 4.30 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Personal Outcomes-Positive Construct Construct

Mean

Standard Deviation (δ δ) 0.529

Cronbach’s Alpha (α α) 0.720

Construct Reliability

Variance Extracted

Personal 4.013 0.710 OutcomesPositive Note: See Appendix K for construct reliability and variance extracted workings.

0.411

4.2.11 The Personal Outcomes-Negative Construct Principal component analysis with varimax rotation was conducted on the four personal outcomes-negative items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 62.64 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.31). Therefore, no items were dropped following this stage of item testing.

Table 4.31 Personal Outcomes-Negative Principal Component Analysis Items Q14. I will be in disgraceful if I do not put into practice what I have learned from the training. Q18. I will be seen as skeptical if I do not put into practice what I have learned from the training. Q19. It is a waste by sending me to the training if I do not put into practice what I have learned from the training. Q22. My job evaluation report will be jeorpardised if I do not put into practice what I have learned from the training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Personal OutcomesNegative 0.80 0.90 0.61 0.84 291 2.51 62.64 0.79

Following this reliability analysis was conducted with these items. Results indicated that all items had item-total correlations above 0.50 except one item (Q19), which displayed a item-total correlation of 0.41 (thus, below the recommended 0.50 cut-off)

118

(Hair et al. 1998) (see Appendix J). An inspection of the inter-item correlation matrix found that (Q19,Q14) displayed an inter-item correlation of 0.28, just slightly below the recommended 0.30 cut-off (Hair et al. 1998) (see Appendix J). However, because the Cronbach’s alpha for this scale was at 0.79, and thus safely above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J), all items were retained for this scale. Next, the measurement model for personal outcomes-negative construct was constructed with the four items as depicted in Figure 4.11 below.

Figure 4.11 Measurement Model: Personal Outcomes-Negative Construct .49

e1

q14 .70

.85

e2

q18

.92 .21

e3

q19 .56

e4

.46

Personal outcomesnegative

.75

q22

Table 4.32 Fit Indices for Personal Outcomes-Negative Construct χ2 value Degrees of freedom p value GFI AGFI

1.844 2 0.398 0.997 0.984

RMSR TLI CFI NFI RMSEA

0.018 1.001 1.000 0.996 0.000

The measurement model for personal outcomes-negative construct produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). 119

The factor loadings for all items were also above the recommended 0.50 cut-off (Bagozzi & Yi 1988) except for one item (q19) which was slightly below the desired cut-off (see Figure 4.11). Further, the fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.32). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.33). Table 4.33 Descriptive Statistics, Cronbach’s alpha, Construct Reliability and Variance Extracted for Personal Outcomes-Negative Construct Construct

Mean

Standard Deviation (δ δ) 0.827

Cronbach’s Alpha (α α) 0.789

Construct Reliability

Personal 3.20 0.809 outcomesnegative Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.528

4.2.12 The Supervisor Sanctions Construct Principal component analysis with varimax rotation was conducted on the four supervisor sanctions items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 72.10 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loading above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.34). Therefore, no items were dropped following this stage of item testing.

120

Table 4.34 Supervisor Sanctions Principal Component Analysis Items

Factor: Supervisor Sanctions 0.85

Q27. My supervisor/manager will say that the application of what I have learned from the training will not bring any benefit. Q28. My supervisor/manager will oppose any application of technique that I have learned from the training. Q31. My supervisor/manager will instruct me with a method that is against to what I have learned from the training. Q32. I will be criticised by my supervisor/manager if I put into practice what I have learned from the training. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

0.88 0.85 0.81 291 2.88 72.10 0.87

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.87, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for supervisor sanctions construct was constructed with the four items as depicted in Figure 4.12 below. Figure 4.12 Measurement Model: Supervisor Sanctions Construct .71

e1

q27 .85

.84

e2

q28

.92 .48

e3

q31 .39

.42

e4

.69

Supervisor sanctions

.63

q32

121

Table 4.35 Fit Indices for Supervisor Sanctions Construct χ2 value Degrees of freedom p value GFI AGFI

0.818 1 0.366 0.999 0.986

RMSR TLI CFI NFI RMSEA

0.005 1.002 1.000 0.999 0.000

The χ2 statistic for the initial measurement model for supervisor sanctions construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination on the modification indices showed that major improvements could be achieved if errors for q31 and q32 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.12). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988). The fit indices showed a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off 0.90 (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.35). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.36). Table 4.36 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Supervisor Sanctions Construct Construct

Mean

Standard Deviation (δ δ) 0.870

Cronbach’s Alpha (α α) 0.870

Construct Reliability

Supervisor 2.233 0.860 sanctions Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.610

122

4.2.13 The Personal Capacity for Transfer Construct Principal component analysis with varimax rotation was conducted on the three personal capacity for transfer items As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 73.32 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.37). Therefore, no items were dropped at this stage. Table 4.37 Personal Capacity for Transfer Principal Component Analysis Items Q15. I am capable to put into practice what I have learned from the training even though I am busy.

Factor: Personal Capacity for Transfer 0.77

Q20. I have a mental capability to put into practice what I have learned from the training in carrying out duties.

0.90

Q21. I have a psychical capability to put into practice what I have learned from the training in carrying out duties.

0.89

Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

291 2.20 73.32 0.80

Following this test, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.80, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for the personal capacity for transfer construct was constructed with the three items as depicted in Figure 4.13 below.

123

Figure 4.13 Measurement Model : Personal Capacity for Transfer Construct .40

e1

q15

.63

.80

e2

q20

.90 .69 .83

e3

Personal capacity for transfer

q21

Table 4.38 Fit Indices for Personal Capacity for Transfer Construct χ2 value Degrees of freedom p value GFI AGFI

2.420 1 0.12 0.994 0.967

RMSR TLI CFI NFI RMSEA

0.020 0.988 0.996 0.993 0.070

The measurement model for personal capacity for transfer produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items were also above the recommended 0.50cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.13). Further, the fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.38). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.39).

124

Table 4.39 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Personal Capacity for Transfer Construct Construct

Mean

Standard Deviation (δ δ) 0.521

Cronbach’s Alpha (α α) 0.803

Construct Reliability

Variance Extracted

Personal 4.047 0.834 capacity for transfer Note: See Appendix K for construct reliability and variance extracted workings.

0.632

4.2.14 The Opportunity to Use Construct Principal component analysis with varimax rotation was conducted on the four opportunity to use items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 66.67 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.40). Therefore, no items were dropped following this stage of item testing. Table 4.40 Opportunity to Use Principal Component Analysis Items Q38. My employer has given me a duty that requires practising what I have learned from the training.

Factor: Opportunity to Use 0.73

Q39. My employer has allocated an enough budget for me to put into practice what I have learned from the training.

0.77

Q43. My employer has allocated the required resources for me to put into practice what I have learned from the training.

0.89

Q44. The required resources allocated by my employer are sufficient for me to put into practice what I have learned from the training.

0.87

Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

291 2.67 66.67 0.83

Following this test reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see

125

Appendix J). Further, the Cronbach’s alpha for this scale was at 0.83, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for opportunity to use construct was constructed with the four items as depicted in Figure 4.14 below.

Figure 4.14 Measurement Model: Opportunity to Use Construct .32

e1

q38 .57

.38

.16

e2

.62

q39 .82

e3

q43 .73

e4

.91

Opportunity to use

.85

q44

Table 4.41 Fit Indices for Opportunity to Use Construct χ2 value Degrees of freedom p value GFI AGFI

3.546 1 0.060 0.994 0.940

RMSR TLI CFI NFI RMSEA

0.010 0.969 0.995 0.993 0.094

The χ2 statistic for the initial measurement model for the opportunity to use construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q38 and q39 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.14).

126

When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.14). The fit indices showed a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off 0.90 (Hair et al. 1998). The RMSR was below the recommended level 0.10, however, the RMSEA was slightly above the recommended 0.08 cut-off (Kline 2005) (see Table 4.41). Construct reliability was above the recommended 0.70 level (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.42). Table 4.42 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Opportunity to Use Construct Construct

Mean

Standard Deviation (δ δ) 0.734

Cronbach’s Alpha (α α) 0.831

Construct Reliability

Opportunity 3.444 0.833 to use Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.565

4.2.15 The Content Validity Construct Principal component analysis with varimax rotation was conducted on the five content validity items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 57.78 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loading above the recommended cut-off 0.35 (Hair et al. 1998) (see Table 4.43). Therefore, no items were dropped at this stage.

127

Table 4.43 Content Validity Principal Component Analysis Items Q33. The training contents are suitable for the duties. Q40. The training contents are related to the need of my duties. Q41. The training contents are important to the need of my duties. Q42. The training contents are something needed for the need of my duties. Q45. The training contents do not fulfil the need for duties*. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha Note: * Negative worded item.

Factor: Content Validity 0.66 0.83 0.87 0.87 0.51 291 2.89 57.78 0.78

Then, reliability analysis was conducted with these items. Result indicated that all items had item-total correlation and inter-item correlation above the recommended cut-off 0.50 and 0.30 respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.83, also above the recommended cut-off 0.70 (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for content validity construct was constructed with the five items as depicted in Figure 4.15.

128

Figure 4.15 Measurement Model: Content Validity Construct .26

e1

q33 .58

e2

.51

q40

.76

.86

e3

.93

q41

.65 .81 -.61

e4

q42

Content validity

.45 .20

e5

q45

Table 4.44 Fit Indices for Content Validity Construct χ2 value Degrees of freedom p value GFI AGFI

7.332 4 0.119 0.990 0.963

RMSR TLI CFI NFI RMSEA

0.012 0.986 0.994 0.988 0.054

The χ2 statistic for the initial measurement model for content validity construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q41 and q45 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.15). When, this was done, the model produced a good fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items also above the recommended cut-off 0.50 (Bagozzi & Yi 1988) except for one item (q45), slightly

129

below the recommended cut-off 0.50 (see Figure 4.15). However, this item was retained because the fit indices showed a good fit with GFI, AGFI, NFI, TLI and CFI above the desired cut-off 0.90 (Hair et al. 1998). The RMSR and RMSEA below the recommended level 0.10 and 0.08 respectively (Kline 2005) (see Table 4.44). The construct reliability above the recommended level 0.70 (Hair et al. 1998) and variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.45). Table 4.45 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Content Validity Construct Construct

Mean

Standard Deviation (δ δ) 0.567

Cronbach’s Alpha (α α) 0.783

Construct Reliability

Variance Extracted

Content 4.029 0.831 Validity Note: See Appendix K for construct reliability and variance extracted workings.

0.512

4.2.16 The Transfer Design Construct Principal component analysis with varimax rotation was conducted on the four transfer design items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 70.62 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loading above the recommended cut-off 0.35 (Hair et al. 1998) (see Table 4.46). Therefore, no items were dropped at this stage. Table 4.46 Transfer Design Principal Component Analysis Items Q34. The training has been delivered systematically. Q35. The training has been delivered effectively. Q36. The training has been delivered in a straightforward approach. Q37. The training has been delivered by using examples that correspond with the need for duties. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Transfer Design 0.77 0.90 0.87 0.81 291 2.83 70.62 0.86

130

Then, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.86, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003). Therefore, all items were retained for this scale. Next, the measurement model for transfer design construct was constructed with the four items as depicted in Figure 4.16 below. Figure 4.16 Measurement Model: Transfer Design Construct .44

e1

q34 .66

.82

e2

q35

.90 .69

e3

q36 .52

e4

.83

Transfer design

.72

q37

Table 4.47 Fit Indices for Transfer Design Construct χ2 value Degrees of freedom P value GFI AGFI

1.649 2 0.438 0.997 0.986

RMSR TLI CFI NFI RMSEA

0.003 1.002 1.000 0.997 0.000

The measurement model for transfer design construct produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al 1998). The factor

131

loadings for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.16). Further, the fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.47). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.48). Table 4.48 Descriptive Statistics, Cronbach’s alpha, Construct Reliability and Variance Extracted for Transfer Design Construct Construct

Mean

Standard Deviation (δ δ) 0.489

Cronbach’s Alpha (α α) 0.857

Construct Reliability

Transfer 4.172 0.862 design Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.613

4.2.17 The Sharing Behaviour Construct Principal component analysis with varimax rotation was conducted on the five sharing behaviour items. As expected, only one factor was extracted (eigenvalue above 1) which accounted for approximately 53.81 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) except one item (Q87), which had a factor loading of 0.14. An inspection found that, this item was a negative worded item and it performed poorly due to its poor wording (see Appendix F for the negative worded item). Thus, it was decided to drop Q87 and the program was re-run. Results indicated that, the percentage of variance increased to 66.95 percent as depicted in Table 4.49 below.

132

Table 4.49 Sharing Behaviour Principal Component Analysis Items Q81. I always being invited to share knowledge and skills that have been learned from the training into the workplace. Q82. I always discuss with my colleague regarding knowledge and skills that have been learned from the training into the workplace. Q83. I always share knowledge and skills that have been learned from the training. Q84. I share knowledge and skills that have been learned from the training into the workplace in the past. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Sharing Behaviour 0.76 0.87 0.84 0.80 291 2.68 66.95 0.83

Following this test, a reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.83, also above the 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, the remaining four items were retained for this scale. Next, the measurement model for sharing behaviour construct was constructed with the four items as depicted in Figure 4.17 below.

Figure 4.17 Measurement Model: Sharing Behaviour Construct .41

e1

q81 .64

.73

e2

q82

.86 .63

e3

q83 .48

e4

.80

Sharing behaviour

.69

q84

133

Table 4.50 Fit Indices for Sharing Behaviour Construct χ2 value Degrees of freedom p value GFI AGFI

5.728 2 0.057 0.991 0.953

RMSR TLI CFI NFI RMSEA

0.010 0.975 0.992 0.987 0.080

The measurement model for the sharing behaviour construct produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loading for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.17). Further, the fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR was below the recommended level 0.10 while RMSEA was on the margin of the recommended level 0.08 (Kline 2005) (see Table 4.50). Construct reliability was above the recommended 0.70 level (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.51). Table 4.51 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Sharing Behaviour Construct Construct

Mean

Standard Deviation (δ δ) 0.583

Cronbach’s Alpha (α α) 0.826

Construct Reliability

Sharing 3.944 0.837 behaviour Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.566

4.2.18 The Intention to Share Construct Principal component analysis with varimax rotation was conducted on the four intention to share items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 70.36 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998).

134

Further, all items had factor loadings above the recommended cut-off 0.35 (Hair et al. 1998) (see Table 4.52). Therefore, no items were dropped at this stage. Table 4.52 Intention to Share Principal Component Analysis Items Q79. I will try to share knowledge and skills that have been learned from the training into the workplace. Q80. I plan to share knowledge and skills that have been learned from the training into the workplace. Q85. I hope to share knowledge and skills that have been learned from the training into the workplace. Q86. I decide to share knowledge and skills that have been learned from the training into the workplace. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor 1: Intention to Share 0.79 0.85 0.85 0.86 291 2.81 70.36 0.86

Following this, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.86, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for intention to share construct was constructed with the four items as depicted in Figure 4.18 below.

135

Figure 4.18 Measurement Model: Intention to Share Construct .30

e1

q79 .54

.41

.56

e2

q80

.64 .75

e3

q85 .85

e4

.86

Intention to share

.92

q86

Table 4.53 Fit Indices for Intention to Share Construct χ2 value Degrees of freedom p value GFI AGFI

1.662 1 0.197 0.997 0.972

RMSR TLI CFI NFI RMSEA

0.002 0.994 0.999 0.997 0.048

The χ2 statistic for the initial measurement model for content validity was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q79 and q80 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.18). When, this was done, the model produced an acceptable fit to the data with a nonsignificant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.18). The fit indices also showed a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al.

136

1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.53). Construct reliability was above the recommended 0.70 level (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.54). Table 4.54 Descriptive Statistics, Cronbach’s alpha, Construct Reliability and Variance Extracted for Intention to Share Construct Construct

Mean

Standard Deviation (δ δ) 0.459

Cronbach’s Alpha (α α) 0.859

Construct Reliability

Intention to 4.165 0.836 share Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.572

4.2.19 The Attitude Toward Knowledge Sharing Construct Principal component analysis with varimax rotation was conducted on the five attitude toward knowledge sharing items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 65.02 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998). It was found that one item (Q74) had a factor loading of 0.39, and thus was above the desired cut-off. However, its loading was not consistent with the loading of other items (Q67,0.89; Q68,0.92; Q69,0.91; Q72,0.80). An inspection found that this item (Q74) was another negatively worded item and it performed poorly due to its poor wording (see Appendix F for the negative worded item). Thus, it was decided to drop Q74 and the program was re-run. Results indicated that the percentage of variance increased to approximately 78.57 percent as depicted in Table 4.55 below.

137

Table 4.55 Attitude Toward Knowledge Sharing Principle Component Analysis Items

Q67. Sharing of knowledge and skills that have during training are good. Q68. Sharing of knowledge and skills that have during training can improve job performance. Q69. Sharing of knowledge and skills that have during training can bring benefits. Q72. Sharing of knowledge and skills that have during training are advantageous. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

been learned

Factor: Attitude Toward Knowledge Sharing 0.90

been learned

0.93

been learned

0.91

been learned

0.80 291 3.14 78.57 0.91

Following this, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.91, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, the remaining four items were retained for this scale. Next, the measurement model for the attitude toward knowledge sharing construct was constructed with the four items as depicted in Figure 4.19 below. Figure 4.19 Measurement Model: Attitude Toward Knowledge Sharing Construct .76

e1

q67 .87

.86

e2

q68

.93 .78

e3

q69 .49

e4

.88 .70

Attitude toward knowledge sharing

q72

138

Table 4.56 Fit Indices for Attitude Toward Knowledge Sharing Construct χ2 value Degrees of freedom p value GFI AGFI

4.175 2 0.124 0.993 0.964

RMSR TLI CFI NFI RMSEA

0.004 0.992 0.997 0.995 0.061

The measurement model for attitude toward knowledge sharing construct produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50 cutoff, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.19). Further, the fit indices also showed support for a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR was below the recommended level 0.10 while RMSEA just reached the recommended level 0.08 (Kline 2005) (see Table 4.56). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.57). Table 4.57 Descriptive Statistics, Cronbach’s alpha, Construct Reliability and Variance Extracted for Attitude Toward Knowledge Sharing Construct Construct

Mean

Standard Deviation (δ δ) 0.498

Cronbach’s Alpha (α α) 0.908

Construct Reliability

Attitude 4.365 0.911 toward knowledge sharing Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.722

139

4.2.20 The Subjective Norm Toward Knowledge Sharing Construct Principal component analysis with varimax rotation was conducted on the four subjective norm toward knowledge sharing items. As expected, only one factor was extracted (eigenvalue above 1), which accounted for approximately 62.99 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cutoff (Hair et al. 1998) (see Table 4.58). Therefore, no items were dropped at this stage. Table 4.58 Subjective Norm Toward Knowledge Sharing Principal Component Analysis Items

Q70. My supervisor/manager thinks that I should share knowledge and skills that have been learned from the training into the workplace. Q71. My head department thinks that I should share knowledge and skills that have been learned from the training into the workplace. Q73. Those who are closed to me think that I should share knowledge and skills that have been learned from the training into the workplace. Q75. Those who are of the same mind with me think that I should share knowledge and skills that have been learned from the training into the workplace. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor 1: Subjective Norm Toward Knowledge Sharing 0.88 0.85 0.80 0.62 291 2.52 62.99 0.79

Following this, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998). Only one item formed the exception: (Q75). This item had an item-total correlation of 0.43, slightly below the recommended 0.50 cut-off (see Appendix J). However, it was decided to retain this item because the Cronbach’s alpha for this scale was 0.79, and thus above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J).

140

Next, the measurement model for subjective norm toward knowledge sharing construct was constructed with the four items as depicted in Figure 4.20 below. Figure 4.20 Measurement Model: Subjective Norm Toward Knowledge Sharing Construct .86

e1

q70 .93

.72

e2

q71

.85 .36

e3

q73 .14

.34

e4

.60 .37

Subjective norm toward knowledge sharing

q75

Table 4.59 Fit Indices for Subjective Norm Toward Knowledge Sharing Construct χ2 value Degrees of freedom p value GFI AGFI

0.665 1 0.415 0.999 0.989

RMSR TLI CFI NFI RMSEA

0.004 1.004 1.000 0.999 0.000

The χ2 statistic for the initial measurement model for the subjective norm toward knowledge sharing construct was significant (p=0.000), indicating that the model did not adequately account for the observed covariation among the items (Hair et al. 1998). Examination of the modification indices showed that major improvements could be achieved if errors for q73 and q75 were allowed to correlate (Arbuckle & Wothe 1999; Joreskog & Sorbom 1993) (see Figure 4.20). When, this was done, the model produced an acceptable fit to the data with a nonsignificant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all

141

items were also above the recommended 0.50 cut-off (Bagozzi & Yi 1988) except for one item (q75), which had a factor loading of 0.37 (see Figure 4.20). However, this item was retained because it showed a good fit with GFI, AGFI, NFI, TLI and CFI above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended level 0.10 and 0.08 respectively (Kline 2005) (see Table 4.59). The construct reliability above the recommended level 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.60). Table 4.60 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Subjective Norm Toward Knowledge Sharing Construct Construct

Mean

Standard Deviation (δ δ) 0.599

Cronbach’s Alpha (α α) 0.793

Construct Reliability

Subjective 3.979 0.798 norm toward knowledge sharing Note: See Appendix K for construct reliability and variance extracted workings.

Variance Extracted 0.521

4.2.21 The Perceived Behavioural Control Construct Principal component analysis with varimax rotation was conducted on the three perceived behavioural control toward knowledge sharing items. As expected, only one factor was extracted (eigenvalue above 1) which accounted for approximately 69.45 percent of the variance, confirming the unidimensionality of this construct (De Vellis 2003; Hair et al. 1998). Further, all items had factor loadings above the recommended 0.35 cut-off (Hair et al. 1998) (see Table 4.61). Therefore, no items were dropped at this stage.

142

Table 4.61 Perceived Behavioural Control Toward Knowledge Sharing Principal Component Analysis Items

Q76. I can share knowledge and skills that have been learned from the training into the workplace without any enforcement. Q77. I can share knowledge and skills that have been learned from the training into the workplace at any time. Q78. I can still share knowledge and skills that have been learned from the training into the workplace even though I am busy. Number of cases Eigenvalue Percentage of Variance Cronbach’s alpha

Factor: Perceived Behavioural Control Toward Knowledge Sharing 0.81 0.89 0.80 291 2.08 69.45 0.77

Following this, reliability analysis was conducted with these items. Results indicated that all items had item-total correlations and inter-item correlations above the recommended cut-offs: 0.50 and 0.30, respectively (Hair et al. 1998) (see Appendix J). Further, the Cronbach’s alpha for this scale was at 0.77, also above the recommended 0.70 cut-off (Carmines & Zeller 1979; De Vellis 2003) (see Appendix J). Therefore, all items were retained for this scale. Next, the measurement model for perceived behavioural control toward knowledge sharing construct was constructed with the three items as depicted in Figure 4.21 below.

143

Figure 4.21 Measurement Model: Perceived Behavioural Control Toward Knowledge Sharing Construct .4 5

e1

q76

.6 7

.7 8

e2

q77

.88 .4 8 .7 0

e3

P erceived behavioural control

q78

Table 4.62 Fit Indices for Perceived Behavioural Control Toward Knowledge Sharing Construct χ2 value Degrees of freedom p value GFI AGFI

2.686 1 0.101 0.994 0.963

RMSR TLI CFI NFI RMSEA

0.019 0.980 0.993 0.990 0.076

The measurement model for perceived behavioural control toward knowledge sharing produced a good model fit to the data with a non-significant χ2 statistic (p>0.05), indicating that the actual and predicted input matrices were not statistically different (Hair et al. 1998). The factor loadings for all items were also above the recommended 0.50 cut-off, indicating adequate individual item reliability (Bagozzi & Yi 1988) (see Figure 4.21). Further, the fit indices showed support for a good fit with GFI, AGFI, NFI, TLI and CFI all above the desired 0.90 cut-off (Hair et al. 1998). The RMSR and RMSEA were below the recommended levels of 0.10 and 0.08, respectively (Kline 2005) (see Table 4.62). Construct reliability was above the recommended level of 0.70 (Hair et al. 1998) and the variance extracted by the construct was larger than the variance extracted due to measurement error (above 0.50), providing support for reliability and validity of this construct (Fornell & Larcker 1981) (see Table 4.63).

144

Table 4.63 Descriptive Statistics, Cronbach’s Alpha, Construct Reliability and Variance Extracted for Perceived Behavioural Control Toward Knowledge Sharing Construct Construct

Mean

Standard Deviation (δ δ) 0.547

Cronbach’s Alpha (α α) 0.771

Construct Reliability

Variance Extracted

Perceived 4.095 0.797 behavioural control toward knowledge sharing Note: See Appendix K for construct reliability and variance extracted workings.

0.571

4.3 Discussion This chapter described the assessment of validity and reliability for each construct in this study using exploratory factor analysis (principal component analysis), confirmatory factor analysis (structural equation modelling) and Cronbach’s alpha. In exploratory factor analysis, principal component analysis with varimax rotation indicated that all constructs under examination extracted only one factor (eigenvalue above 1), confirming the unidimensionality of all constructs under study. The variance extracted by all constructs ranged from 55.26 percent to 82.11 percent. The personal-outcomes positive construct recorded the lowest while openness to change recorded the highest variance. According to Hinkin (1995), it is important that the scale used for construct measurement generates sufficient variance for statistical analysis. In this thesis, a five-point Likert-type scale that ranged from ‘1=Strongly Disagree’ to ‘5=Strongly Agree’ was utilised. As the variance extracted by all constructs were above 50 percent, this implies the suitability of this type of scale to demonstrate higher variance when applied in the context of this study. The Cronbach’s alpha reliabilities for all constructs were above the desired cut off 0.70, and ranged from 0.72 to 0.93. Only two constructs displayed Cronbach’s alpha reliabilities slightly above the desired cut off 0.70 while others above 0.80. The two constructs were personal outcomes-positive (0.72) and learner readiness (0.73). The Cronbach’s alphas for these constructs were lower than other constructs because several items in these constructs were not consistent in measuring what they suppose 145

to measure due to poor wordings. For example, two items belonged to learner readiness construct: ‘I know that this training is good for me’ and ‘I applied for this training on my own’ were not consistent with the other three items for the construct. A similar problem occurred for the personal outcomes-positive construct. Two of the items: ‘My work will be rewarded if I put into practice what I have learned’ and ‘I will get a good result from the job evaluation report whenever I put into practice what I have learned from the training’ were not consistent with the other two items for the construct. These problematic items were retained in this thesis because the overall Cronbach’s alpha for the constructs were above the desired cut-off 0.70. However, future research should replace these items in order to improve the consistency reliability of the scales. Further, three negative worded items also performed poorly. The three negative worded items belonged to the constructs: transfer effort-performance expectations, attitude toward knowledge sharing and sharing behaviour. In this thesis, the researcher used negative worded items in order to avoid agreement bias (Churchill 1979; De Vellis 2003; Spector 1992). However, it was found that the items created confusion for the respondents instead of attempting to avoid agreement bias. In this thesis, the researcher had no choice but to drop the three items in order to increase the reliability and validity of the three constructs. In the transfer of training literature, two studies were found using negative worded items and reported similar problems as occurred in this thesis (Chiaburu & Marinova 2005; Holton et al. 2000). Not surprisingly, several researchers opposed the use of negative worded items suggesting that they may introduce confusion rather than precision (Jackson, Wall, Martin & Davids 1993). Although the guidelines are available for the wording of negative worded items (De Vellis 2003), the researcher found that it was not as easy as suggested. Therefore, an important lesson learned in this thesis is to take particular care in properly wording negative items and to test them using different samples to see how they perform before they can be accepted and used.

146

Also important to highlight in the discussion is the number of items for each construct. In this thesis, the number of items ranged from three to five. Two constructs had three items each; 14 constructs had four items each; and five constructs had five items. Although Hinkin (1995) suggested five or six items for each construct to provide adequate internal consistency reliability, the researcher found that three items for each construct proved sufficient. Several other researchers had also suggested that a measure should have at least three items for content adequacy (Cook et al. 1989). In this thesis, the two constructs having three items showed good internal consistency reliability where the Cronbach’s alpha were 0.80 (personal capacity for transfer) and 0.77 (perceived behavioural control toward knowledge sharing), both above the desired cut-off 0.70 (De Vellis 2003). Moreover, the researcher felt that having more than three items for each construct would make the questionnaire lengthy and could then be difficult to obtain co-operation from the respondents. The researcher encountered this experience during the data collection. The questionnaire comprised 87 items in total. Most respondents expressed concern about the length of the questionnaire and they took about 30 to 40 minutes to complete it. The researcher also received about 67 incomplete questionnaires (more than one page incomplete). Arguably, if three good items were used for each construct, the length of the questionnaire could have been reduced to 63 items and consequently less time would have been taken by respondents to answer the questionnaire. Perhaps then, the number of incomplete questionnaire could also have been reduced. The use of confirmatory factor analysis in this thesis is to filter the best items that can represent the constructs under study (Chin 1988). Because some items were allowed to correlate, construct assessment in this chapter should be regarded as exploratory rather than confirmatory (Chin 1998; Hurley et al. 1997). In this thesis, several items belonging to 12 constructs were allowed to correlate. One example is the performance-self efficacy construct. Two items belong to this construct were allowed to correlate due to item redundancy. The two items were: ‘I am confident to increase my job performance’ and ‘I have the capabilities to increase my job performance’.

147

These two items were redundant and therefore errors for these items were allowed to correlate. When errors for these items were correlated, the construct reliability and variance extracted for performance-self efficacy construct improved. The variance extracted by all constructs was also above the desired 0.50 cut-off, except for three constructs: learner readiness (0.455), performance-outcomes expectations (0.478) and personal outcomes-positive (0.411) constructs. Each of these constructs emerged slightly below the desired 0.50 cut-off but principle component analysis with varimax rotation revealed that the variance extracted were 55.36 percent, 61.46 percent and 55.26 percent respectively, clearly above the 50 percent. The construct reliability for these constructs were also above the desired cut-off 0.70, indicating that despite the slightly lower variance extractions, these constructs possessed adequate reliability and validity. Overall, the researcher was satisfied that all measures possessed good psychometric properties (validity and reliability) to investigate the relationships hypothesised in this thesis. As described above, in exploratory factor analysis, principal component analysis with varimax rotation indicated that all constructs under examination extracted only one factor (eigenvalue above 1), confirming the unidimensionality of all constructs under study. The Cronbach’s alphas for all constructs above the desired cut-off 0.70. In confirmatory factor analysis, by exception of learner readiness, performance-outcomes expectations and personal outcomes-positive, all constructs above the desired cut-off 0.70 and 0.50 for construct reliability and variance extracted respectively.

4.4 Summary This chapter continued the discussion on methodology (Chapter 3) through a presentation of the results of construct assessment using the techniques of exploratory factor analysis (principal component analysis), confirmatory factor analysis (structural equation modelling) and Cronbach’s alpha. A total of 21 constructs were tested for this study of which three items were rejected. The rejected items were all 148

negatively worded questions. Their removal significantly increased the validity and reliability of the constructs under investigation. The chapter confirms that all constructs actually measure what they intended to measure and all display good psychometric properties (validity and reliability) at this point of study to investigate the relationships hypothesised in this thesis. Chapter 5 moves to a detailed discussion of the hypotheses testing of the 10 hypotheses driving this study along with the results and their implications for transfer of training.

149

CHAPTER 5 HYPOTHESIS TESTING: RESULTS AND DISCUSSION

5.1 Introduction Chapter 3 described the research framework and the methodology chosen to examine the relationships hypothesised. Then, Chapter 4 continued the methodological discussion through an examination of tests conducted to ensure the reliability and validity of all constructs under study. Chapter 4 detailed the use of exploratory factor analysis (principal component analysis), Cronbach’s alpha and confirmatory factor analysis (structural equation modelling) for each construct used in the survey instrument. Chapters 5 and 6 discuss the findings from the research questions and related hypotheses. This Chapter considers in detail the findings from research questions one and two. The Chapter commences with an overview of the types of respondents to the survey before moving to test the relevant hypotheses.

5.2 Distribution of Respondents Before describing hypothesis testing and presenting the results, it is useful to consider the distribution of respondents by training types, gender, age, education level, work experience and position of employment. This demographic data assists in better understanding the types of trainees in the Malaysian public sector and from that, the phenomenon of transfer of training at the level of the individual trainee. This information is presented below.

150

5.2.1 Distribution of Respondents by Training Types As discussed in Chapter 2, transfer of training research has been quite sparse in Malaysia despite increasing pressure on the public sector to deliver a return on investment in terms of improved training transfer and to better link training to improvements in job performance. To assist in rectifying this research gap, a cross sectional study of the Malaysian public sector was used in this thesis with a survey questionnaire as the main instrument for data collection. The sample chosen consisted of government employees attending training at the National Institute of Public Administration, a central training organisation for government employees in Malaysia. Data were collected from respondents who had attended three types of training:

general

training

(for

example,

quality

report

writing),

management/leadership training (for example, human resource management) and computer training (for example, visual basic). Appendix E provides further details of the training types. Table 5.1 below presents the distribution of respondents by training types. Table 5.1 Distribution of Respondents by Training Types Training Types General Training

n 180

Percent 61.9

Management/Leadership Training Computer Training

71

24.4

40

13.7

291

100

Total

From Table 5.1, it can be seen that most respondents attended general training (61.9 percent) with 24.4 percent of respondents attending management/leadership training. Only 13.7 percent of respondents had attended computer training.

151

5.2.2 Distribution of Respondents by Gender The information regarding gender indicated that male and female respondents were almost equally distributed. Table 5.2 below presents the distribution of respondents by gender. Table 5.2 Distribution of Respondents by Gender Male

Gender

Female Total

n 152

Percent 52.2

139

47.8

291

100

5.2.3 Distribution of Respondents by Age The information regarding respondents’ ages indicated that most who attended training were 30 years of age or below (49.5 percent). This was followed by respondents in the age bracket between 31 and 40 years (30.6 percent). Senior respondents aged 41 and above, represented the lowest group of trainees (19.9 percent). Table 5.3 below presents the distribution of respondents by age. Table 5.3 Distribution of Respondents by Age Age 30 years and below

N 144

Percent 49.5

Between 31 and 40 years

89

30.6

41 years and above

58

19.9

291

100

Total

152

5.2.4 Distribution of Respondents by Level of Education The information regarding respondents’ levels of education indicated that those who held a bachelor degree and those who had less than a bachelor degree were almost equally distributed. Table 5.4 below presents the distribution of respondents by their level of education. Table 5.4 Distribution of Respondents by Level of Education Level of Education Less than Bachelor Degree

n 137

Percent 47.1

Bachelor above

154

52.9

291

100

Degree Total

and

5.2.5 Distribution of Respondents by Work Experience The information regarding respondents work experience indicated that most had work experience of less than 5 years (43.7 percent). This was followed by respondents who had more than 10 years work experience (33.3 percent). Respondents who had work experience between 5 to 10 years recorded the lowest attendance (23 percent). Table 5.5 below presents the distribution of respondents by their work experience. Table 5.5 Distribution of Respondents by Work Experience Work Experience Less than 5 years

n 127

Percent 43.7

Between 5 to 10 years

67

23.0

Above 10 years

97

33.3

Total

291

100

153

5.2.6

Distribution

of

Respondents

by

Position

of

Employment The information regarding respondents’ positions of employment indicated that those from support groups and management groups were almost equally distributed. Table 5.6 below presents the distribution of respondents by their position of employment. Table 5.6 Distribution of Respondents by Position of Employment Status of Employment Management Group

n 147

Percent 50.5

Support Group

144

49.5

291

100

Total

This section overviewed the types of respondents and the training undertaken in this study. In general, almost equal numbers of men and women participated in a range of training dominated by general training. There were also roughly equal numbers of respondents who occupied either support positions or managerial positions and this is likely reflected in the fact that approximately equal numbers of degree qualified and less than degree qualified individuals participated in the survey. Training attendance was prevalent amongst trainees with less than 5 years of work experience. The researcher believes that, it would be more instructive to have compared the demographics of the sample with the demographics of the population from which it was drawn. However, the data regarding the demographics of the population (by gender; age; level of education; work experience; position of employment) from which it was drawn was not obtained from the training provider. The researcher found that, although the registration process at the training centre has been computerised, however, demographics data by gender; age; level of education; work

154

experience and position of employment cannot be sorted by the system at the training centre.

5.3 Hypothesis Testing In Chapter 1, six research questions were introduced which formed the basis of the enquiry into transfer of training in a Malaysian public sector context. A total of 10 hypotheses were then constructed to assist in answering the research questions. In this Chapter, Hypothesis 1 (H1) and Hypothesis 2 (H2) belonging to research questions one and two respectively are tested and the results are presented below.

5.3.1 Hypothesis Testing for Research Question One Research Question one asked: Which of these transfer of training variables: •

motivation to transfer;



secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use);



enabling (content validity, transfer design); and



TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean

score

across

different

training

types

(general

training,

management/leadership training, computer training)?

155

One hypothesis was developed to answer research question one: Hypothesis 1 (H1): The following transfer of training variables: motivation to transfer; secondary influences (performance-self efficacy, learner readiness); expected utility (transfer effort-performance expectations, performance-outcomes expectations); transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions); ability (personal capacity for transfer, opportunity to use); enabling (content validity, transfer design) and TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their

mean

scores

across

different

training

types

(general

training,

management/leadership training, computer training). H1 Testing: As described in the methodology chapter (Chapter 3), MANOVA was selected to test H1. The 21 variables listed above were used as the dependent variables while the training types (general training, management/leadership training, computer training) were used as independent variables (Coakes & Steed 2003; Pallant 2005). The ratio of the largest group (general training) to the smallest group (computer training) was 4.5 (180 ÷ 40). According to Hair et al. (1998), if the ratio of the largest group to the smallest group is more than 1.5, then a test for equality of covariance (Box’s Test) is needed. The Box’s Test significant value should be larger than 0.001 (non significant difference) to indicate the equality of covariance across groups, in this case general training, management/leadership training and computer training (Coakes & Steed 2003; Pallant 2005). In the present study, Box’s Test showed a significant difference (p = 0.000). However, because Box’s Test is not a robust test (Harris 1985; Yamnill

156

2001), it was decided to run MANOVA and the results were interpreted as described below. The results of the multivariate test showed a statistically significance difference. The Wilk’s Lambda was significant at the 0.000 level (Wilks’ Lambda = 0.708; F = 2.405) indicating that there were significant differences among the variables in terms of their means across the three different training types. Then, the test of betweensubject effect output box was checked in order to identify which variables differed. The results indicated that of the 21 variables, two were significantly different (p < 0.002) across the three training types. These two variables were feedback and supervisor sanctions, both transfer climate variables (see Appendix L1). The Levene’s test of equality of error variances showed that feedback (p = 0.014) and supervisor sanctions (p = 0.307) were not significant, indicating that these two variables did not violate the assumption of homogeneity of variance. Further examination on the mean scores of the two variables that differed, found that the respondents from computer training rated higher than the respondents from general training and management/leadership training on feedback scale (M = 3.969 versus 3.857 and 3.592 respectively) and supervisor sanctions scale (M = 2.856 versus 2.113 and 2.187 respectively). Therefore, H1 was supported for these two variables.

5.3.2 The relationship between the transfer of training variables and the type of training undertaken The findings for the first research question revealed that only two variables were significantly different across the three training types. The two variables were feedback and supervisor sanctions. Both are transfer climate variables and they are considered below:

157

The Effect of Feedback Feedback was found to be significantly different across the three types of training. Trainees from computer training rated feedback higher than trainees from general and management/leadership training (M = 3.969 versus 3.857 and 3.592 respectively). As described in Chapter 1, feedback represents the formal and informal indicators from an organisation about an individual’s job performance. The results suggest that trainees from computer training are more likely to receive feedback regarding their job performance than trainees from general and management/leadership training. However, the fact that the overall mean score across the three training types was 3.808, indicates that feedback is at best operating at a moderate level only. Supervisors and managers in the Malaysian public sector need to provide greater feedback to trainees regarding their job performance as a result of attending training if they want to see transfer of training occur. This has also been observed in a range of studies finding that an individual’s performance will be further enhanced if feedback on job performance is provided (Clarke 2002; Reber & Wallin 1984; Tracey et al. 1995) (see also Chapter 3, section 2.5.1) It is likely that providing trainees with feedback could motivate them to transfer training to the job. Chapter 1 defined motivation to transfer as the trainees’ desire to use the knowledge and skills mastered in the training program on the job. In this thesis, motivation to transfer was not significantly different across the training types. All trainees indicated strong levels of motivation to transfer (M = 4.367, 4.310 and 4.231 for general management/leadership and computer training, respectively). Although motivation to transfer operated at a strong level, the overall mean score of 4.334, indicated that continuous effort should be made by trainers to ensure that this level is maintained and improved. As feedback operated at a moderate level, it is argued that providing trainees with greater feedback could stimulate and improve their motivation to transfer too. This issue is taken up in Chapter 7 when discussing the importance of feedback in motivating a trainee to transfer training.

158

The Effect of Supervisor Sanctions The ratings for Supervisor sanctions were found to be significantly different across the three training types. Trainees from computer training rated this variable higher than trainees from general and management/leadership training (M = 2.856a versus 2.113 and 2.187 respectively). Chapter 1 described supervisor sanctions as the extent to which individuals perceive negative responses from supervisors/managers when applying skills learned in training. The results suggest that trainees from computer training likely face a stronger level of sanction, or negative response from their supervisors when applying training on the job than those trainees from the other training types. Whilst trainees across all training types generally rated supervisor sanctions quite low, it is nevertheless a pertinent message to supervisors in the Malaysian public sector that they need adopt a more facilitative approach if they want to see transfer of training to occur. Similarly, supervisors should decrease their opposition if they want their trainees to have greater motivation to transfer training to the job. Again, this point is taken up in Chapter 7.

The Effect of Secondary Influence Variables In this thesis, performance-self efficacy and learner readiness were not significantly different across the training types. All trainees indicated strong levels of performance-self

efficacy

(M

=

4.168,

4.194

and

4.119

for

general,

management/leadership and computer training respectively) and learner readiness (M = 4.410, 4.485, 4.365 for general, management/leadership and computer training respectively). Chapter 1 defined performance-self efficacy as trainees’ beliefs that they are able to change their performance when they want to and learner readiness as the extent to which trainees are prepared to enter and participate in training. The results of this study suggest that trainees across all the three training types feel ready to participate in training and they have the confidence to increase their job a

For supervisor sanctions, the average mean score between 1.0 to 2.0 indicates a less likely of supervisor sanctions; between 2.0 and 2.5 indicates a moderate level of supervisor sanctions and above 2.5 indicates a strong level of supervisor sanctions.

159

performance at will. The results indicate that as employees in the Malaysian public sector possess learner readiness and performance-self efficacy they are motivated to transfer training to the job. This finding provides the building blocks from which a focus on developing programs which enhance transfer of training can be built. This issue is taken up in Chapter 7. The Effect of Expected Utility Variables The expected utility variables: transfer effort-performance expectations and performance-outcomes expectations were not significantly different across the training types. Chapter 1 defined transfer effort-performance expectations as the expectation that the effort devoted to transferring learning will lead to changes in job performance. All trainees across the three types of training rated this variable strongly (M = 4.054). In other words, trainees reported strong levels of expectation that their efforts would convert to greater job performance. These results also suggest the usefulness of Vroom’s (1964) expectancy theory in understanding trainees’ expectations in a workplace training context. Briefly, as described in Chapter 2, Vroom postulated that individuals will be more motivated when they believe that their effort invested in the training program will result in mastery of the training content. In contrast, trainees across the three training types rated performance-outcomes expectations relatively poorly (M = 3.385). Chapter 1 defined performance-outcomes expectations as the expectation that changes in job performance will lead to valued outcomes. In other words, trainees were indicating their low expectations that changes in job performance would necessarily be rewarded. In this study, it was consistently found that expectations regarding intrinsic outcomes are more important to trainees than are extrinsic outcomes. For instance, consider the item related to an expected intrinsic outcome such as “I expect that I will be more entrusted if my job performance increased”. This item was agreed to by 79 percent of the respondents whereas an item related to expected extrinsic outcomes such as “I expect that I will be

160

promoted if my job performance increased” was rated as uncertain by 51 percent of the respondents. Together, these results fit with the fact that employers in the Malaysian public sector do not provide extrinsic rewards to trainees when they have improved their performance as a result of attending training. Consequently, trainees expect their rewards will be of a more intrinsic nature (for instance, being more entrusted, satisfied or empowered) rather than extrinsic (for instance, being promoted). The latter instance is particularly true in the Malaysian public sector where job promotion is not given simply on the completion of training, even though job performance may increase as a result of attending training. This point is further taken up in Chapter 7 in a discussion on the implications of this study for improving transfer of training in the Malaysian public sector. The Effect Transfer Climate Variables Transfer climate consists of important environmental elements which influence motivation to transfer (Holton 1996; Seyler et al. 1998). With the exceptions of feedback and supervisor sanctions, all other climate variables were not significantly different across the training types. For instance, in the analysis of the transfer climate variables, peer support and supervisor support were found not significantly different across the three training types. Chapter 1 defined peer and supervisor support as reinforcing and supporting the use of learning to the job. Trainees rated these variables (M = 3.589 for peer support and 3.791 for supervisor support). The findings suggest that peers and supervisors should improve the level of support they provide trainees back on the job if motivation to transfer is to be enhanced. On the other hand, all trainees across the three types of training rated openness to change and personal outcomes-negative relatively poorly (M = 2.745b and 3.200 b

For openness to change, the average mean score between 1.0 to 2.0 indicates an openness to change; between 2.0 and 2.5 indicates a moderate level of openness to change and above 2.5 indicates a poor level of openness to change.

161

respectively). Chapter 1 defined openness to change as the extent to which prevailing group norms are perceived by trainees as discouraging the use of skills and knowledge acquired in training while personal outcomes-negative is defined as the extent to which trainees believe that not applying skills and knowledge learned in training will lead to negative outcomes. The low findings may imply that rigidity of teamwork or reticence to accept new ideas or to make any changes in the way the work is normally done is a problem in the workplace. Further, trainees do not have a strong belief that negative outcomes, such as warnings or punishment, would happen to them if they failed to apply training on the job. Interestingly, trainees across the three types of training rated personal outcomespositive strongly (M = 4.013). Chapter 1 defined personal outcomes-positive as the degree to which applying training on the job leads to outcomes that are positive for the trainees. It was found in this study that trainees expect to receive rewards which are not based on monetary or other external sources but on intrinsic factors such as personal satisfaction. As with responses to feedback, described above, items related to intrinsic personal outcomes-positive sentiments such as “I will work with more confidence if I put into practice what I have learned from the training” generated agreement by 95 percent of the respondents. This finding indicates that confidence itself is a reward for trainees. Several studies conducted in the public sector in the United States have also suggested that intrinsic rewards proved to be more important than extrinsic rewards (Bates 2001; Kontoghiorghes 2001). Other transfer climate variables such as feedback and supervisor sanctions were significantly different across the three training types and the effect of these variables was discussed above. Overall, despite the exception provided in the findings on the personal outcomes-positive variable, transfer climate in the Malaysian public sector could not, at this stage, be described as being at an appropriate level to promote training transfer. Although trainees across the three training types have a strong level of performance-self efficacy, learner readiness and motivation to transfer, other transfer climate variables such as feedback, peer support, supervisor support,

162

openness to change, personal outcomes-negative may inhibit optimal levels of training transfer as these were not at appropriate levels. Further, trainees across the three training types reported the likelihood of facing sanctions from their supervisors. Together, these results suggest that HRD managers in the Malaysian public sector could take measures to improve the transfer climate if greater levels of transfer is to occur. This point is taken up in Chapter 7. The Effect of Enabling Variables Both the enabling variables: content validity and transfer design were not significantly different across the training types. Trainees across rated content validity and transfer design highly (M = 4.029 and 4.172 respectively). Chapter 1 defined content validity as the extent to which trainees judge training content to accurately reflect job requirements and transfer design as the degree to which training has been designed and delivered to give trainees the ability to transfer learning to the job and training instructions match job requirements. These results seem to imply that the training content across the training types does not suffer from a disconnection with trainees’ job requirements and indeed, that training instructions match trainees’ job requirements. In other words, there is a strong coherence between the requirements of the job and the training to perform those required tasks. This is a strong finding in terms of the Malaysian government’s investment in public sector training as it demonstrates an appropriate match at the level of content and job specificity of this training. The implication of this as a factor which may enhance transfer of training is discussed in Chapter 7. The Effect of Ability Variables Both the ability variables: personal capacity for transfer and opportunity to use were not significantly different across the three training types. Trainees across the three training types rated personal capacity for transfer highly (M = 4.048) and opportunity to use, relatively poorly (M = 3.444). Chapter 1 defined personal capacity for transfer as the extent to which individuals have the time, energy and mental space in their work lives to make changes required to transfer learning to the

163

job while opportunity to use was defined as the extent to which trainees are provided with or obtain resources and tasks on the job. The findings suggest that trainees have the capacity to transfer training but they do not have the opportunity to use training on the job. It could be that trainees are not given the tasks that require them to use their training on the job or it could be that the required resources (for instance, computer equipment or finance) are not available and this inhibits their opportunity to use training on the job. The findings suggest that these two factors may act against each other in the Malaysian public sector in terms of transfer of training. This point is further taken up in Chapter 7 in a discussion on creating the conditions, which better enhance transfer of training. The Effect Sharing Variables (TPB) It is interesting to highlight in this discussion that the TPB variables were not significantly different across the three training types. Trainees across the three training types rated the sharing variables highly. For instance, knowledge sharing behaviour (the degree to which trainees share the learned knowledge and skills in the workplace) averaged M = 3.944. Similarly, Intention to share (the degree to which trainees are willing to share the learned knowledge and skills in the workplace) had a mean score of M = 4.165; attitude (trainees positive or negative evaluations on sharing the learned knowledge and skills in the workplace) was rated at M = 4.365. The mean score for subjective norm (perceived social pressure to share the learned knowledge and skills in the workplace) was M = 3.979; and perceived behavioural control (perceived ease or difficulty of sharing the learned knowledge and skills in the workplace) was M = 4.095. These are important and interesting findings because in the current training transfer literature (as outlined in Chapter 2), researchers have examined secondary influence variables, expected utility variables, transfer climate variables, enabling variables and ability variables (Chen 2003; Holton et al. 2003; Yamnill 2001) but none had included sharing behaviour across different types of training. Thus, this thesis makes a contribution to the current transfer literature, and HRD practices in the Malaysian public sector in particular, by linking motivation to transfer to sharing behaviour. In this study, trainees across the three types of training

164

reported strong levels of motivation to transfer and sharing behaviour. This point is further taken up in Chapter 7. This section described the findings from research question one demonstrating that trainees across the training types have strong levels of motivation to transfer, learner readiness, performance-self efficacy, transfer effort-performance expectations, personal outcomes-positive and personal capacity for transfer. The training programs themselves were rated strongly in terms of content validity and transfer design. Nevertheless, it appears that trainees are less likely to transfer training to the job because the transfer climate variables: feedback, peer support, supervisor support and personal outcomes-negative are relatively moderate. Further, upon the completion of training, trainees across the three training types face problems with openness to change, supervisor sanctions and lack of opportunity to use training on the job. The findings from research question one could help HRD managers in the Malaysian public sector (and possibly elsewhere) to intervene and improve those variables that are weak in order to promote training transfer. This topic is taken up in Chapter 7. The findings from research question one also showed that sharing behaviour was stable and rated highly across the training types. This is an indicator that sharing behaviour can be a potential factor, which promotes training transfer because it can be generalised across training types. This Chapter now moves to a consideration of research question two.

5.3.3 Hypothesis Testing for Research Question Two Research question two asked: Which of these transfer of training variables: •

motivation to transfer;



secondary influences (performance-self efficacy, learner readiness);



expected utility (transfer effort-performance expectations, performanceoutcomes expectations);

165



transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions);



ability (personal capacity for transfer, opportunity to use),



enabling (content validity, transfer design); and



TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across trainees’ demographics (gender, age, level of education, work experience, position of employment)?

Hypothesis 2 (H2): The following transfer of training variables: motivation to transfer; secondary influences (performance-self efficacy, learner readiness); expected utility (transfer effort-performance expectations, performance-outcomes expectations); transfer climate (feedback, peer support, supervisor support, openness to change, personal outcomes-positive, personal outcomes-negative, supervisor sanctions); ability (personal capacity for transfer, opportunity to use); enabling (content validity, transfer design) and TPB (sharing behaviour, intention to share, attitude toward knowledge sharing, subjective norms toward knowledge sharing, perceived behavioural control toward knowledge sharing) are significantly different in terms of their mean score across trainees’ demographics (gender, age, level of education, work experience, position of employment). MANOVA was used to test H2. The 21 variables listed above were used as the dependent variables while trainees’ demographics (gender, age, level of education, work experience, position of employment) were used as independent variables (Coakes & Steed 2003; Pallant 2005). As described in Chapter 3, these six independent variables were analysed separately, one by one, using one-way MANOVA, starting with gender and then followed by: age, level of education, work experience and position of employment. The results are presented below.

166

Gender As indicated above, Gender was used as the independent variable while the 21 variables in H2 were the dependent variables (Coakes & Steed 2003; Pallant 2005). The ratio of the largest group (male) to the smallest group (female) was 1.09 (152 ÷ 139). According to Hair et al. (1998), if the ratio of the largest group to the smallest group is less than 1.5, then a test for equality of covariance (Box’s Test) is not needed. Then, MANOVA was run and the results were interpreted. The results of the multivariate test showed significant difference. The Wilk’s Lambda was significant (p